
%A D. Sumanaweera
%A L. Allison
%A A. Konagurthu
%T Statistical compression of protein sequences and inference of marginal
probability landscapes over competing alignments using finite state models
and Dirichlet priors
%J Bioinformatics
%V 35
%N 14
%P i360–i369
%M JUL
%D 2019
%O ISMB/ECCB, Basel, .ch, July 2019
%K jrnl, conf, MolBio, c2019, c201x, c20xx, zz0719, LAllison, ArunK,
bioinformatics, sequence, alignment, landscape, DPA, homology, protein,
proteins, probability, probabilistic, Bayesian, information, MML, MDL,
ISMB, ECCB
%X "The information criterion of minimum message length (MML) provides a
powerful statistical framework for inductive reasoning from observed data. We
apply MML to the problem of protein sequence comparison using finite state
models with Dirichlet distributions. The resulting framework allows us to
supersede the ad hoc cost functions commonly used in the field, by
systematically addressing the problem of arbitrariness in alignment
parameters, and the disconnect between substitution scores and gap costs.
Furthermore, our framework enables the generation of marginal probability
landscapes over all possible alignment hypotheses, with potential to
facilitate the users to simultaneously rationalize and assess competing
alignment relationships between protein sequences, beyond simply reporting a
single (best) alignment. We demonstrate the performance of our program on
benchmarks containing distantly related protein sequences."
 [doi:10.1093/bioinformatics/btz368]['19].
Also see [protein].
%A A. S. Konagurthu
%A R. Subramanian
%A L. Allison
%A D. Abramson
%A M. Garcia de la Banda
%A P. J. Stuckey
%A A. M. Lesk
%T Informationtheoretic inference of an optimal dictionary of protein
supersecondary structures
%B Protein Supersecondary Structures
%I Springer
%P 123131
%M APR
%D 2019
%K chapter, c2019, c201x, c20xx, zz0519, LAllison, ArunK, AMLesk,
bioinformatics, protein, structure, proteins, supersecondary, fold, tableau,
dictionary, library, MML, minimum message length, mdl, AI, inference
%X "We recently developed an unsupervised Bayesian inference methodology to
automatically infer a dictionary of protein supersecondary structures
(Subramanian et al., IEEE data compression conference proceedings (DCC),
340349, 2017). Specifically, this methodology uses the informationtheoretic
framework of minimum message length (MML) criterion for hypothesis selection
(Wallace, Statistical and inductive inference by minimum message length,
Springer Science & Business Media, New York, 2005). The best dictionary of
supersecondary structures is the one that yields the most (lossless)
compression on the source collection of folding patterns represented as
tableaux (matrix representations that capture the essence of protein folding
patterns (Lesk, J Mol Graph. 13:159–164, 1995). This book chapter outlines
our MML methodology for inferring the supersecondary structure dictionary.
The inferred dictionary is available at [www(click)][/4/2019]."
 [doi:10.1007/9781493991617_6]['19].
(Also see [protein].)
%A L. Allison
%T Coding Ockham's Razor
%I Springer
%D 2018
%K book, text, c2018, c201x, c20xx, zz0618, LAllison, Occam, Ockham razor, MDL,
minimum message length, description, MML, AIC, BIC, statistics, probability,
likelihood, inductive inference, Bayesian, artificial intelligence,
algorithm, decision, tree, Dtree, nml, mmld, kdd, clustering, snob,
mixture model, modelling, iris data, regression, stats, machine learning,
II, AI, 27 club, 27club, musician, graph, graphs, graphModel, information,
compression, model, code, random, entropy, maths, stats, L4ref, L5ref
%X "... inductive inference using the minimum message length (MML) principle and
a computer. It is accompanied by a library of software to help an
applications programmer, student or researcher in the fields of data analysis
or machine learning to write computer programs of this kind. ..."
 [doi:10.1007/9783319764337].
uk us isbn:3319764322; uk us isbn13:9783319764320.
(Also see the [mmlist].)
%A D. Sumanaweera
%A L. Allison
%A A. S. Konagurthu
%T The bits between proteins
%J Data Compression Conference (DCC)
%I IEEE
%W Snowbird, Utah, USA
%M MAR
%D 2018
%K conf, DCC, DCC2018, MolBio, c2018, c201x, c20xx, zz0418, bioinformatics,
LAllison, ArunK, Dinithi, DCC, DCC2018, protein, sequence, alignment, DPA,
algorithm, minimum message length, MDL, MML, AIC, homology, information,
similarity
%X "Comparison of protein sequences via alignment is an important routine in
modern biological studies. Although the technologies for aligning proteins
are mature, the current state of the art continues to be plagued by many
shortcomings, chiefly due to the reliance on: (i) naive objective functions,
(ii) fixed substitution scores independent of the sequences being considered,
(iii) arbitrary choices for gap costs, and (iv) reporting, often, one
optimal alignment without a way to recognise other competing sequence
alignments. Here, we address these shortcomings by applying the
compressionbased Minimum Message Length (MML) inference framework to the
protein sequence alignment problem. This grounds the problem in statistical
learning theory, handles directly the complexityvsfit tradeoff without
ad hoc gap costs, allows unsupervised inference of all the statistical
parameters, and permits the visualization and exploration of competing
sequence alignment landscape."
 [more],
[doi:10.1109/DCC.2018.00026]['18].
(Also see [protein].)
%A R. Subramanian
%A L. Allison
%A P. J. Stuckey
%A M. Garcia de la Banda
%A D. Abramson
%A A. M. Lesk
%A A. S. Konagurthu
%T Statistical compression of protein folding patterns for inference of
recurrent substructural themes
%J Data Compression Conf. (DCC)
%I IEEE
%W Snowbird, Utah, USA
%P 340349
%M APR
%D 2017
%K conf, DCC, MolBio, c2017, c201x, c20xx, zz0517, protein, tertiary, 3D,
super secondary, structure, motif, motifs, pattern, blocks, library,
discovery, description, aic, MDL, minimum message length, MML,
LAllison, AMLesk, ArunK
%X "Computational analyses of the growing corpus of threedimensional (3D)
structures of proteins have revealed a limited set of recurrent substructural
themes, termed supersecondary structures. Knowledge of supersecondary
structures is important for the study of protein evolution and for the
modeling of proteins with unknown structures. Characterizing a comprehensive
dictionary of these supersecondary structures has been an unanswered
computational challenge in protein structural studies. This paper presents an
unsupervised method for learning such a comprehensive dictionary using the
statistical framework of lossless compression on a database comprised of
concise geometric representations of protein 3D folding patterns. The best
dictionary is defined as the one that yields the most compression of the
database. Here we describe the inference methodology and the statistical
models used to estimate the encoding lengths. An interactive website for this
dictionary is available ..."
 [more]
& [doi:10.1109/DCC.2017.46]['17].
(Also see [protein].)
%A J. H. Collier
%A L. Allison
%A A. M. Lesk
%A P. J. Stuckey
%A M. Garcia de la Banda
%A A. S. Konagurthu
%T Statistical inference of protein structural alignments using information and
compression
%J J. Bioinformatics
%I OUP
%V 33
%N 1
%P 10051013
%M APR
%D 2017
%O bioRxiv, June 2016
%K jrnl, OUP, MolBio, c2017, c201x, c20xx, zz0317, protein, alignment,
tertiary structure, 3D, information, MML, MMLigner, software,
JHCollier, ArunK, LAllison, AMLesk, AIC, bic, mdl
%X "... present here a statistical framework for the precise inference of
structural alignments, built on the Bayesian and informationtheoretic
principle of Minimum Message Length (MML). The quality of any alignment is
measured by its explanatory power—the amount of lossless compression achieved
to explain the protein coordinates using that alignment. ..."
 [doi:10.1093/bioinformatics/btw757][2017] (online January 2017),
[bioRxiv][6/2016],
[more].
(Also see MMLigner@[LCB][2016].)
%A M. Duc Cao
%A L. Allison
%A T. I. Dix
%A M. Boden
%T Robust estimation of evolutionary distances with information theory
%J Mol. Biol. Evol.
%V 33
%N 5
%P 13491357
%M MAY
%D 2016
%K jrnl, MBE, MolBio, c2016, c201x, c20xx, zz0816, bioinformatics, evolutionary,
phylogenetic, distance, trees, sequences, information, maximum, parsimony,
DNA, likelihood, MML, MDL, AIC, Minh Duc Cao, LAllison, TIDix, MBoden
%X "Methods for measuring genetic dist. in phylogenetics are known to be
sensitive to the evol. model assumed. However, there is a lack of established
methodology to accommodate the tradeoff between incorporating suff. biol.
reality & avoiding model overfitting. [&] as trad. methods measure dists.
based on the observed # of substitutions, they tend to underestimate dist.
between diverged seqs. due to backward & parallel subsns. Various techniques
were proposed to correct this, but they lack the robustness [v.] sequs. that
are distantly related & of unequal base freqs.. ... we present a novel
genetic dist. est. based on inf.theory that overcomes the above two hurdles.
Instead of examining the obs. # of subsns, [it] ests. genetic dist. using
Shannon's mutual information. This naturally provides an effective framework
for balancing model complexity & goodness of fit. Our dist. est. is shown to
be approx. linear to elapsed time & hence is less sensitive to the divergence
of seq. data & compositional biased seqs.. Using extensive simulation data,
we show that our method 1) consistently reconstructs more accurate phylogeny
topologies than existing methods, 2) is robust in extreme condns such as
diverged phylogenies, unequal base freqs. data, & heterogeneous mutation
patterns, & 3) scales well with large phylogenies." [online Feb 2016.]
 [doi:10.1093/molbev/msw019]['16].
[Also search for: MolBio evolutionary MML].
%A A. S. Konagurthu
%A P. Kasarapu
%A L. Allison
%A J. H. Collier
%A A. M. Arthur
%T On sufficient statistics of leastsquares superposition of vector sets
%J J. Comp. Biol.
%V 22
%N 6
%P 487497
%M MAY
%D 2015
%K jrnl, JCB, MolBio, bioinformatics, ArunK, LAllison, AMLesk, JHCollier,
c2015, c201x, c20xx, zz0116, 3D, point set, tertiary, protein, structure,
structural alignment, superposition, match, matching, stats
%X "The problem of superposition of two corr. vector sets by minimizing their
sumofsquares error under orthogonal transformation ... can be solved
exactly using an alg. whose time complexity grows linearly with the # of
correspondences. ... particularly in studies involving macromolecular
structs.. ... formally derives a set of suff.stats. for the leastsquares
superposition problem. These s. are additive. This permits a highly efficient
(const. time) computation of superpositions (& s.stats.) of vector sets that
are composed from its constituent v.sets under addition or deletion op.,
where the s.stats. of the constituent sets are already known (that is, [they]
have been previously superposed). ... a drastic improvement in the run time
of the methods that commonly superpose v.sets under addition or deletion
ops., where previously these ops. were carried out ab initio (ignoring the
s.stats.). ... demonstrate the improvement our work offers in the context of
protein structural alignment programs that assemble a reliable structural
alignment from wellfitting (substructural) fragment pairs. A C++ library for
this task is available online under an opensource license."
 [doi:10.1089/cmb.2014.0154]['16].
(Based on the 2014 RECOMB paper.)
%A P. Kasarapu
%A L. Allison
%T Minimum message length estimation of mixtures of multivariate Gaussian and
von MisesFisher distributions
%J Machine Learning
%V 100
%N 2
%P 333378
%M MAR
%D 2015
%O ECML / PKDD, Porto, Portugal, 711 Sept. 2015
%K jrnl, c2015, c201x, c20xx, zz0415, AI, II, Parthan, LAllison, MDL, AIC,
MML, minimum message length, mixture modelling, model, clustering, vonMises,
multivariate, vMF, Fisher, protein, structure, description, classification,
Bayesian, directional, probability distribution, mmld, Figuerdo
%X "Mixture modelling involves explaining some observed evidence using a
combination of probability distributions. The crux of the problem is the
inference of an optimal number of mixture components and their corr.
parameters. This paper discusses unsupervised learning of mixture models
using the Bayesian Minimum Message Length (MML) criterion. To demonstrate the
effectiveness of search & inference of mixture parameters using the proposed
approach, we select two key probability distributions, each handling
fundamentally different types of data: the multivariate Gaussian distribution
to address mixture modelling of data distributed in Euclidean space, & the
multivariate von MisesFisher (vMF) distribution to address mixture modelling
of directional data distributed on a unit hypersphere. The key contributions
of this paper, in addition to the general search & inference methodology,
include the derivation of MML expressions for encoding the data using
multivariate Gaussian & von MisesFisher distributions, & the analytical
derivation of the MML estimates of the parameters of the two distributions.
Our approach is tested on simulated & real world data sets. For instance, we
infer vMF mixtures that concisely explain experimentally determined 3D
protein conformations, providing an effective null model description of
protein structures that is central to many inference problems in structural
bioinformatics. The experimental results demonstrate that the performance of
our proposed search & inference method along with the encoding schemes
improve on the state of the art mixture modelling techniques."
 [doi:10.1007/s1099401554930]['15],
[more].
%A F. Petitjean
%A L. Allison
%A G. Webb
%T A statistically efficient and scalable method for loglinear analysis of
highdimensional data
%J ICDM
%I IEEE
%W Shenzhen (.cn)
%M DEC
%D 2014
%K proc, conf, ICDM, DM2018, c2014, c201x, c20xx, zz1214, data mining, models,
graphical model, log linear, chordal graph, chordalysis, MML, AIC, mdl, chi2,
chi squared, II, AI, maths, stats, decomposable, FPj, LAllison, GWebb, Monash
%X "Loglinear analysis is the primary statistical approach to discovering
conditional dependencies between the variables of a dataset. A good
loglinear analysis method requires both high precision & statistical
efficiency. High precision means that the risk of false discoveries should be
kept very low. Statistical efficiency means that the method should discover
actual associations with as few samples as possible. Classical approaches to
loglinear analysis make use of chi^2 tests to control this balance between
quality & complexity. We present an informationtheoretic approach to
loglinear analysis. We show that our approach 1) requires significantly
fewer samples to discover the true associations than statistical approaches
 statistical efficiency  2) controls for the risk of false discoveries as
well as statistical approaches  high precision  and 3) can perform the
discovery on datasets with hundreds of variables on a standard desktop
computer  computational efficiency."
 [more],
[doi:10.1109/ICDM.2014.23]['14].
%A J. Collier
%A L. Allison
%A A. Lesk
%A M. Garcia de La Banda
%A A. Konagurthu
%T A new statistical framework to assess structural alignment quality using
information compression
%J ECCB
%W Strasbourg
%M SEP
%D 2014
%K conf, ECCB 14, MolBio, c2014, c201x, c20xx, zz0914, LAllison, ArunK, AMLesk,
JHCollier, protein, 3D, similar, structure, alignment, match, MML, MDL, AIC,
complexity, bioinformatics, 13th Euro, Conf, Comp, Biology, I value, Ivalue
%X "... proposes a new statistical framework to assess structural alignment
quality and significance based on lossless information compression. This is
a radical departure from the traditional approach of formulating scoring
functions. It links the structural alignment problem to the general class of
statistical inductive inference problems, solved using the
informationtheoretic criterion of minimum message length. Based on this, we
developed an efficient and reliable measure of structural alignment quality,
Ivalue. The performance of Ivalue is demonstrated in comparison with a
number of popular scoring functions, on a large collection of competing
alignments. Our analysis shows that Ivalue provides a rigorous and reliable
quantification of structural alignment quality, addressing a major gap in
the field."
 [doi:10.1093/bioinformatics/btu460]['14],
[more].
%A L. Allison
%T On the complexity of graphs (networks) by information content, and
conditional (mutual) information given other graphs
%I Clayton School of the Faculty of Info. Tech., Monash Uni.
%R 2014/277
%P 14
%M JUN
%D 2014
%K c2014, c201x, c20xx, zz0614, TR, complex, compress, compression, graph,
network, model, graphModel, complexity, simple, minimum message length, MML,
random, description, AI, information content, AIC, MDL, mutual, similarity,
viagra, cialis, xanax, valium, structure, chemistry, SMILES, organic, maths,
chemical compounds, molecules, probability, likelihood, LAllison
%X '... concerns the information content of a graph, optionally conditional on
one or more background, "common knowledge" graphs. It describes an algorithm
to estimate this information content, and includes some examples based on
chemical compounds.'
 [more],
pdf@[mon@sh]['14].
%A A. S. Konagurthu
%A P. Kasarapu
%A L. Allison
%A J. H. Collier
%A A. M. Lesk
%T On sufficient statistics of leastsquares superposition of vector sets
%J RECOMB
%I SpringerVerlag
%S LNCS/LNBI
%V 8394
%M APR
%P 144159
%D 2014
%K conf, RECOMB, MolBio, c2014, c201x, c20xx, zz0514, ArunK, LAllison, AMLesk,
JHCollier, bioinformatics, RECOMB18, protein, structure, alignment,
least squares, RMS, error, 3D, match, matching, additive, orthogonal, rigid,
vector set, Kearsley, algorithm
%X "Superposition by orthogonal transformation of vector sets by minimizing the
leastsquares error is a fundamental task in many areas of science, notably
in structural molecular biology. Its widespread use for structural analyses
is facilitated by exact solns of this problem, computable in linear time.
However, in several of these analyses it is common to invoke this
superposition routine a very large number of times, often operating (through
addition or deletion) on previously superposed vector sets. This paper
derives a set of sufficient statistics for the leastsquares orthogonal
transformation problem. These sufficient statistics are additive. This
property allows for the superposition parameters (rotation, translation, &
root mean square deviation) to be computable as constant time updates from
the statistics of partial solutions. We demonstrate that this results in a
massive speed up in the computational effort, when compared to the method
that recomputes superpositions ab initio . Among others, protein structural
alignment algorithms stand to benefit from our results."
 [doi:10.1007/9783319052694_11]['14],
[more].
%A A. S. Konagurthu
%A L. Allison
%A D. Abramson
%A P. J. Stuckey
%A A. M. Lesk
%T How precise are reported protein coordinate data?
%J Acta Cryst.
%V D70
%N 3
%P 904906
%M MAR
%D 2014
%K jrnl, MolBio, c2014, c201x, c20xx, zz0314, protein, 3D, tertiary, structure,
precision, accuracy, PDB, ArunK, LAllison, AMLesk
%X "Atomic coordinates in the Worldwide Protein Data Bank (wwPDB) are generally
reported to greater precision than the experimental structure determinations
have actually achieved. By using information theory & data compression to
study the compressibility of protein atomic coordinates, it is possible to
quantify the amount of randomness in the coordinate data & thereby to
determine the realistic precision of the reported coordinates. On avg., the
value of each C_alpha coordinate in a set of selected p.structures solved at
a variety of resolutions is good to about 0.1A."
 [doi:10.1107/S1399004713031787]['14],
[more].
%A A. S. Konagurthu
%A A. M. Lesk
%A D. Abramson
%A P. J. Stuckey
%A L. Allison
%T Statistical inference of protein "LEGO bricks"
%J ICDM
%M DEC
%D 2013
%K conf, ICDM, ICDM13, MolBio, c2013, c201x, c20xx, zz1213, LAllison, ArunK,
AMLesk, protein, tertiary, 3D, MML, structure, structures, recurrent,
structural, motifs, backbone, folds, MDL, library, dictionary, fragment,
blocks, fragments, bioinformatics, data mining
%X "Proteins are biomolecules of life. They fold into a great variety of
threedimensional (3D) shapes. Underlying these folding patterns are many
recurrent structural fragments or building blocks (analogous to "LEGO(r)
bricks"). This paper reports an innovative statistical inference approach to
discover a comprehensive dictionary of protein structural building blocks
from a large corpus of experimentally determined protein structures. Our
approach is built on the Bayesian and informationtheoretic criterion of
minimum message length [MML]. To the best of our knowledge, this work is the
first systematic and rigorous treatment of a very important data mining
problem that arises in the crossdisciplinary area of structural
bioinformatics. The quality of the dictionary we find is demonstrated by its
explanatory power  any protein within the corpus of known 3D structures can
be dissected into successive regions assigned to fragments from this
dictionary. This induces a novel onedimensional representation of three
dimensional protein folding patterns, suitable for application of the rich
repertoire of characterstring processing algorithms, for rapid
identification of folding patterns of newly determined structures. This paper
presents the details of the methodology used to infer the dictionary of
building blocks, and is supported by illustrative examples to demonstrate its
effectiveness and utility."
 [doi:10.1109/ICDM.2013.73]['14],
[more], and
1310.1462@[arXiv]['13].
%A A. S. Konagurthu
%A A. M. Lesk
%A L. Allison
%T Minimum message length inference of secondary structure from protein
coordinate data
%J J. Bioinformatics
%I OUP
%V 28
%N 12
%P i97i105
%M JUN
%D 2012
%O ISMB, Long Beach
%K conf, ISMB12, MolBio, c2012, c201x, c20xx, zz0612, LAllison, ArunK, AMLesk,
SST, bioinformatics, protein, secondary structure, DSSP, assignment, helix,
extended strand, sheet, coil, mmld, fold, MML, MDL, model
%X "Motivation: Secondary structure underpins the folding pattern and
architecture of most proteins. Accurate assignment of the SS elts is
therefore an important problem. Although many approx. solns of the SS
assignment problem exist, the statement of the problem has resisted a
consistent & math. rigorous defn. A variety of comparative studies have
highlighted major disagreements in the way the available methods define &
assign SS to coord.data.
Results: We report a new method to infer SS based on the Bayesian method of
Minimum Message Length (MML) inference. It treats assignments of SS as
hypotheses that explain the given coord.data. The method seeks to maximise
the joint probability of a hypothesis & the data. There is a natural null
hypothesis & any assignment that cannot better it is unacceptable. We
developed a program SST based on this approach & compared it to popular
programs such as DSSP & STRIDE amongst others. Our evaln suggests that SST
gives reliable assignments even on low resolution structures."
 [doi:10.1093/bioinformatics/bts223]['12].
More: [www]['12].
%A E. Makalic
%A L. Allison
%T MMLD inference of multilayer perceptrons
%J Solomonoff 85th Memorial Conference
%W Melbourne
%I SpringerVerlag
%S LNCS / LNAI
%V 7070
%P 261272
%M DEC
%D 2011
%K conf, Ray, c2011, c201x, c20xx, AI, II, Enes, enesm, MML, mmld, LAllison,
perceptron, NN, ANN, neural network, information, MDL, complexity
%X "A multilayer perceptron comprising a single hidden layer of neurons with
sigmoidal transfer fns can approx. any computable fn to arbitrary accuracy.
The size of the hidden layer dictates the approximation capability of the
multilayer perceptron & automatically determining a suitable network size for
a given data set is an interesting question. ... considers the problem of
inferring the size of multilayer perceptron networks with the MMLD model
selection criterion which is based on the MML principle. The two main
contributions of the paper are: (1) a new model seln criterion for inference
of fullyconnected multilayer perceptrons in regression problems, & (2) an
efficient alg. for computing MMLDtype codelengths in mathematically
challenging model classes. Empirical performance of the new alg. is
demonstrated on artificially generated & real data sets."
 [doi:10.1007/9783642449581_20]['14]; uk us isbn13:9783642449574.
(Also see [MML].)
%A A. S. Konagurthu
%A L. Allison
%A P. J. Stuckey
%A A. M. Lesk
%T Piecewise linear approximation of protein structures using the principle of
minimum message length
%J J. Bioinformatics
%V 27
%N 13
%P i43i51
%M JUL
%D 2011
%K conf, MolBio, MML, c2011, c201x, c20xx, zz0711, ISMB, LAllison, ArunK,
AMLesk, protein, fold, cartoon, description, ribbon diagram, structure,
segmentation, minimum message length, MDL, information theoretic
%X "Simple & concise representations of proteinfolding patterns provide
powerful abstractions for visualizations, comparisons, classifications,
searching & aligning structural data. Structures are often abstracted by
replacing standard secondary structural features  that is, helices & strands
of sheet  by vectors or linear segments. Relying solely on std secondary
structure may result in a sig. loss of structural information. Further,
traditional methods of simplification crucially depend on the consistency &
accuracy of external methods to assign SS to protein coord.data. Although
many methods exist automatically to identify SS, the impreciseness of
definitions, along with errors & inconsistencies in experimental structure
data, drastically limit their applicability to generate reliable simplified
representations, especially for structural comparison.
This article introduces a mathematically rigorous alg. to delineate protein
structure using the elegant statistical & inductive inference framework of
minimum message length (MML). Our method generates consistent & statistically
robust piecewise linear explanations of protein coordinate data, resulting in
a powerful & concise representation of the structure. The delineation is
completely independent of the approaches of using hydrogenbonding patterns
or inspecting local substructural geometry that the current methods use.
Indeed, as is common with applications of the MML criterion, this method is
free of parameters & thresholds, in striking contrast to the existing
programs which are often beset by them.
The analysis of results over a large number of proteins suggests that the
method produces consistent delineation of structures that encompasses, among
others, the segments corresponding to standard secondary structure."
 [doi:10.1093/bioinformatics/btr240]['11].
(Also see [more].)
%A M. D. Cao
%A T. I. Dix
%A L. Allison
%T A biological compression model and its applications
%B Software Tools and Algorithms for Biological Systems
%E H. R. Arabnia
%E Q.N. Tran
%I SpringerVerlag
%S AEMB (Advances in Experimental Medicine and Biology)
%V 696
%P 657666
%M APR
%D 2011
%K MolBio, AEMB, book chapter, c2011, c201x, zz0411, c20xx, LAllison, TIDix,
Minh Duc Cao, bioinformatics, entropy, compression, DNA, protein, sequence,
data, compress, MML, MDL
%X "A biological compression model, expert model, is presented which is superior
to existing compression algorithms in both compression performance and speed.
The model is able to compress whole eukaryotic genomes. Most importantly, the
model provides a framework for knowledge discovery from biological data. It
can be used for repeat element discovery, sequence alignment and phylogenetic
analysis. We demonstrate that the model can handle statistically biased
sequences and distantly related sequences where conventional knowledge
discovery tools often fail."
 [more],
chapter 67 [doi:10.1007/9781441970466_67]['11],
in: hb us$204; uk us isbn:1441970452; uk us isbn13:9781441970459.
%A M. D. Cao
%A T. I. Dix
%A L. Allison
%T A genome alignment algorithm based on compression
%J BMC Bioinformatics
%V 11
%N 1
%P 599
%M DEC
%D 2010
%K jrnl, eJrnl, MolBio, c2010, c201x, c20xx, Minh Duc Cao, LAllison, TIDix,
highly accessed paper, DNA, whole genome alignment, zz0111, local alignment,
sequence, long, compress, compression, MML, MDL, BIC, poa, XM, expert model,
XMAligner
%X "... Since genomic sequences carry genetic info., this article proposes that
the info. content of each nucleotide in a posn should be considered in
sequence alignment. An info.theoretic approach for pairwise genome local
alignment, namely XMAligner, is presented. Instead of comparing sequences at
the character level, XMAligner considers a pair of nucleotides from two
seqs. to be related if their mutual info. in context is significant. The
info.content of nucleotides in sequences is measured by a lossless
compression technique. ... Experiments on both simulated data & real data
show that XMAligner is superior to conventional methods especially on
distantly related seqs. & statistically biased data. XMAligner can align
seqs. of eukaryote genome size with only a modest hardware requirement. ..."
 [more],
[doi:10.1186/1471210511599][12/'10] (BMC 17/2/12 mail: 3372 accesses),
21159205@[pubmed][2/'11].
%A M. B. Dale
%A L. Allison
%A P. E. R. Dale
%T Model selection using Minimal Message Length: an example using pollen data
%J Community Ecology
%V 11
%N 2
%P 187201
%M DEC
%D 2010
%K jrnl, c2010, c201x, c20xx, LAllison, MML, mdl, pollen, data, segmentation,
model, clustering, time series, timeSeries, plant, ecology, vegetation,
cores, sediment
%X "... use of the minimum message length criterion [for] evaluating alternative
models of data when the samples are serially ordered in space & implicitly
in time. Much data from vegetation studies can be arranged in a sequence &
in such cases the user may elect to constrain the clustering by zones, in
preference to an unconstrained clustering. We use the MML principle to
determine if such a choice provides an effective model of the data. Pollen
data provide a suitably organised set of samples, but have other properties
which make it desirable to examine several different models for the
distribution of palynomorphs within the clusters. The results suggest that
zonation is not a particularly preferred model since it captures only a small
part of the patterns present. It represents a user expectation regarding the
nature of variation in the data & results in some patterns being neglected.
By using unconstrained clustering within zones, we can recover some of this
overlooked pattern. We then examine other evidence for the nature of change
in vegetation & finally discuss the usefulness of the MML as a guiding
principle in model choice & its relationship to other possible criteria."
 [doi:10.1556/ComEc.11.2010.2.7]['11].
(Also see [IP].)
%A A. Konagurthu
%A L. Allison
%A T. Conway
%A B. BeresfordSmith
%A J. Zobel
%T Design of an efficient outofcore read alignment algorithm
%J WABI
%I SpringerVerlag
%S LNCS/LNBI
%V 6293
%P 189201
%M SEP
%D 2010
%K wShop, MolBio, c2010, c201x, c20xx, zz1010, WABI, WABI10, nicta, LAllison,
Syzygy, ArunK, Algorithms in Bioinformatics, short read, reads, NGS, align,
mapping, next generation, DNA, sequencing, algorithm
%X "New genome sequencing technologies are poised to enter the sequencing
landscape with significantly higher throughput of read data produced at
unprecedented speeds & lower costs per run. However, current inmemory
methods to align a set of reads to one or more reference genomes are
illequipped to handle the expected growth of readthroughput from newer
technologies. ... reports the design of a new outofcore read mapping alg.,
Syzygy, which can scale to large volumes of read & genome data. The alg. is
designed to run in a constant, userstipulated amount of main memory 
small enough to fit on standard desktops  irrespective of the sizes of read
& genome data. Syzygy achieves a superior spatial localityofreference that
allows all large data structures used in the alg. to be maintained on disk.
We compare our prototype implementation with several popular read alignment
programs. Our results demonstrate clearly that Syzygy can scale to very
large read volumes while using only a fraction of memory in comparison,
without sacrificing performance."
 [more],
[doi:10.1007/9783642152948_16]['10].
(In: uk us isbn:3642152937; uk us isbn13:9783642152931.)
%A M. B. Dale
%A L. Allison
%A P. E. R. Dale
%T A model for correlation within clusters and its use in pollen analysis
%J Community Ecology
%V 11
%N 1
%P 5158
%M JUN
%D 2010
%K jrnl, c2010, c201x, c20xx, LAllison, MML, mdl, pollen, data, plants, ecology,
model, segmentation, clustering, sediments, time
%X "... determine the # of clusters & also whether the model of any cluster is
improved by introducing a factor. ... allows us to seek clusters which
reflect directional changes rather than imposing a zonation constrained by
spatial (& implicitly temporal) position. Minimal message length is a means
of utilising Okham's Razor in inductive analysis. The "best" model is that
which allows most compression of the data, which results in a minimal
message length for the description. Fit to the data is not a sufficient
criterion for choosing models because more complicated models will almost
always fit better. MML combines fit to the data with an encoding of the model
& provides a Bayesian probability criterion as a means of choosing between
models (& classes of model). Applying the analysis to a pollen diagram from
Southern Chile, we find that the introduction of factors does not improve the
overall quality of the mixture model. The solution without axes in any
cluster provides the most parsimonious solution. Examining the cluster with
the best case for a factor to be incorporated in its description shows that
the attributes highly loaded on the axis represent a contrast of herbaceous
vegetation & dominant forests types. This contrast is also found when fitting
the entire population, & in this case the factor solution is the preferred
model. Overall, the cluster solution without factors is much preferred. Thus,
in this case classification is preferred to ordination although more data are
desirable to confirm such a conclusion."
 [doi:10.1556/ComEc.11.2010.1.8][11].
(Also see [IP].)
%A M. D. Cao
%A L. Allison
%A T. I. Dix
%T A distance measure for genome phylogenetic analysis
%J AI09
%I SpringerVerlag
%S LNCS
%V 5866/2009
%P 7180
%M DEC
%D 2009
%K conf, MolBio, c2009, c200x, c20xx, zz1209, Minh Duc Cao, TIDix, AI, II, AI09,
LAllison, NICTA, bioinformatics, MML, MDL, phylogenetic, tree, evolutionary,
XM, trees, distance, measure, matrix, information, maximum parsimony,
maximum likelihood, method, methods, construction, reconstruction, inference,
Plasmodium, Plasmodia
%X "Phylogenetic analyses of species based on single genes or parts of the
genomes are often inconsistent because of factors such as variable rates of
evolution & horizontal gene transfer. ... phylogeny construction from
complete genomes that is less sensitive to such inconsistency. For such long
seqs., construction methods like maximum parsimony & maximum likelihood are
often not possible due to their intensive computational requirement. Another
class of tree construction methods, namely distancebased methods, require a
measure of distances between any 2 genomes. ... presents an information
theoretic measure of genetic distances between genomes based on the
biological compression algorithm 'expert model' [XM]. ... tree of a number
of Plasmodium parasites from their genomes ... successfully construct a
plausible evolutionary tree for the gammaProteobacteria group whose genomes
are known to contain many horizontally transferred genes."
 [doi:10.1007/9783642104398_8]['09],
& [phylogenetics].
(2009; uk us isbn:364210438X; uk us isbn13:9783642104381).
%A J. Bernal
%A T. I. Dix
%A L. Allison
%A A. Bartholomeusz
%A L. Yuen
%T Modelling Hepatitis B virus antiviral therapy and drug resistant mutant
strains
%J ACAL09
%I SpringerVerlag
%S LNCS
%V 5865
%P 159168
%M DEC
%D 2009
%K conf, ACAL, MolBio, c2009, c200x, c20xx, zz1109, TIDix, LAllison, NICTA,
vidrl, Hepatitis B, Hep B, HepB, HBV, bioinformatics, drug resistance, model,
simulation
%X "... an individualbased model of HBV inside an artificial liver, &
associated blood serum, undergoing antiviral therapy. This model aims to
provide insights into the evolution of the HBV quasispecies & the
individual contribution of HBV mutations in the outcome of therapy."
 [doi:10.1007/9783642104275_16]['09].
(2009; uk us isbn:3642104266; uk us isbn13:9783642104268.)
(Also see [Bioinformatics].)
%A M. D. Cao
%A T. I. Dix
%A L. Allison
%T Computing substitution matrices for genomic comparative analysis
%J Advances in Knowledge Discovery and Data Mining
%I SpringerVerlag
%S LNCS
%V 5476 / 2009
%P 647655
%M APR
%D 2009
%K conf, PAKDD, MolBio, c2009, c200x, c20xx, zz0509, bioinformatics, PAKDD09,
PAKDD13, DNA, genome, comparative analysis, comparison, substitution matrix,
PAM, blosum, estimation, alignment, free, homology, MML, MDL, information,
XM, Minh Duc Cao, TIDix, LAllison, (NICTA), malaria genome
%X "Substitution matrices ... and are important for many knowledge discovery
tasks such as phylogenetic analysis and sequence alignment. ... present a
novel algorithm that addresses this by computing a nucleotide substitution
matrix specifically for the two genomes being aligned. ... uses compression
... reconstructs, with high accuracy, the substitution matrix for
synthesised data generated from a known matrix with introduced noise. ...
successfully applied to real data for various malaria parasite genomes,
which have differing phylogenetic distances and composition that lessens the
effectiveness of standard statistical analysis techniques."
pdf@[doi:10.1007/9783642013072_64]['09], and
[substitution matrices].
%A M. D. Cao
%A T. I Dix
%A L. Allison
%T A genome alignment algorithm based on compression
%R 2009/233
%I Faculty of Info. Tech. (Clayton), Monash University
%M JAN
%D 2009
%K TR, TR233, MolBio, c2009, c200x, c20xx, bioinformatics, zz0109, FIT, Monash,
Minh Duc Cao, TIDix, LAllison, data, DNA, sequence, compression, compressed,
compress, MML, MDL, align, strings, information, content, free, XM, XMAligner
%X "Traditional genome alignment methods based on dynamic programming are often
a. computational expensive, b. unable to compare the genomes of distant
species, & c. unable to deal with low information regions. ... information
theoretic approach for pairwise genome local alignment. ... the expert model
aligner, the XMAligner, relies on the expert model compression alg.. To align
2 seqs., XMAligner 1st compresses one sequence to measure the info. content
at each posn in the seq.. Then the seq. is compressed again but this time
with the background knowledge from the other seq. to obtain the conditional
info. content. The info. content & the conditional info. content from the
2 compressions are examined. Similar regions in the compressed seq. should
have the conditional info. content lower than the individual info. content.
... applied to align the genomes of Plasmodium falciparum & P. knowlesi v.
other 3 P. genomes with different levels of diversity. Despite the
differences in nucleotide composition of the reference seqs., the conserved
regions found by XMAligner in 3 alignments are relatively consistent. A
strong correlation was found between the similar regions detected by the
XMAligner & the hypothetical annotation of Plasmodium species. The alignment
results can be integrated into the DNAPlatform for visualisation."
 [abs]['09].
[Also search for: Cao Dix Allison Bioinformatics c2010],
and see [compression]['09].
%A L. Allison
%T Added distributions for use in clustering (mixture modelling), function
models, regression trees, segmentation, and mixed Bayesian networks in
Inductive Programming 1.2
%R 2008/224
%I Faculty of Info. Tech., Monash University, Australia 3800
%M APR
%D 2008
%K IP 1.2, IP1.2, TR 224, TR224, c2008, c200x, c20xx, zz0508, LAllison,
machine learning, Geometric, Poisson, Gaussian, probability, stats, model,
tree, segment, network, FP, MDL, minimum message length, MML, description,
inductive inference, II, Student's tDistribution, tDistribution
%X "Inductive programming is a machine learning paradigm combining functional
programming (FP) with the information theoretic criterion, Minimum Message
Length (MML). IP 1.2 now includes the Geometric and Poisson distributions
over nonnegative integers, and Student's tDistribution over continuous
values, as well as the Multinomial and Normal (Gaussian) distributions from
before. All of these can be used with IP's modeltransformation operators,
and structurelearning algorithms including clustering (mixturemodels),
classification (decision) trees and other regressions, and mixed Bayesian
networks, provided only that the types match between each corresponding
component Model, transformation, structured model, and variable  discrete,
continuous, sequence, multivariate, and so on."
 [Ind.Prog.(IP)] inc. source code,
[Paper.ps].
%A T. I. Dix
%A D. R. Powell
%A L. Allison
%A J. Bernal
%A S. Jaeger
%A L. Stern
%T Comparative analysis of long DNA sequences by per element
information content using different contexts
%J BMC Bioinformatics
%V 8(S2):S10
%M MAY
%D 2007
%K jrnl, ejrnl, MolBio, c2007, c200x, c20xx, zz0407, TIDix, LAllison, sequence,
analysis, data, compression, compress, low, mutual, information content,
simple, complex, DNA, plot, II, MDL, AI, context, description, profile,
signature, pattern discovery, genomic, minimum message length, MML,
Cyanidioschyzon merolae, alga, algae, Plasmodium falciparum, gui, browser
%X "... The paper presents a methodology for exploring long DNA sequences, of
the order of millions of bases, by means of their information content. ..."
 [long DNA],
(@[csse]),
[doi:10.1186/147121058S2S10]['08],
17493248@[pubmed]['11].
%A M. D. Cao
%A T. I. Dix
%A L. Allison
%A C. Mears
%T A simple statistical algorithm for biological sequence compression
%J Data Compression Conf.
%W Snowbird, Utah
%I IEEE
%P 4352
%M MAR
%D 2007
%K conf, MolBio, II, DCC, DCC07, c2007, c200x, c20xx, zz0307, DNA, protein,
minh, minhc, Minh Duc Cao, LAllison, TIDix, MML, MDL, compress, entropy,
information theory, expert model, XM, profile, models, bioinformatics,
dnacompress, pattern discovery
%X "This paper introduces a novel algorithm for biological sequence compression
that makes use of both statistical properties & repetition within sequences.
A panel of experts is maintained to estimate the probability distribution of
the next symbol in the sequence to be encoded. Expert probabilities are
combined to obtain the final distribution. The resulting information sequence
provides insight for further study of the biological sequence. Each symbol is
then encoded by arithmetic coding. Experiments show that our algorithm
outperforms existing compressors on typical DNA & protein sequence datasets
while maintaining a practical running time."
[compression],
[compression],
[reprint.pdf],
 [doi:10.1109/DCC.2007.7]['07].
%A R. T. O'Donnell
%A K. B. Korb
%A L. Allison
%T Causal KL: Evaluating causal discovery
%I Faculty of Info. Tech., Monash University, Clayton, Australia 3800
%R 2007/207
%P 26
%M FEB
%D 2007
%K TR 207, TR207, c2007, c200x, c20xx, LAllison, rodo, KBKorb, AI, II,
causal model, stats, intervention, interventions, Bayesian network, networks,
BN, models, cause, Causal Kullback Leibler distance, KL, CKL, CKL,
Bnet, Bnets
%X "The two most commonly used criteria for assessing causal model discovery
with artificial data are editdistance & KullbackLeibler divergence,
measured from the true model to the learned model. Both of these metrics
maximally reward the true model. However, we argue that they are both
insufficiently discriminating in judging the relative merits of false models.
Edit distance, for example, fails to distinguish between strong & weak
probabilistic dependencies. KL divergence, on the other hand, rewards
equally all statistically equivalent models, regardless of their different
causal claims. We propose an augmented KL divergence, which we call Causal
KL (CKL), which takes into account causal relationships which distinguish
between observationally equivalent models. Results are presented for three
variants of CKL, showing that Causal KL works well in practice."
[pdf]['07],
[abs]['07],
Also see [MML].
%A M. B. Dale
%A L. Allison
%A P. E. R. Dale
%T Segmentation and clustering as complementary sources of information
%J Acta Oecologica
%I Elsevier
%V 31
%N 2
%P 193202
%M MAR
%D 2007
%K jrnl, ecology, LAllison, c2007, c200x, c20xx, segment, cluster, MML, MDL,
II, minimum message length, description, classification, mixture model,
modelling, BIC, ecological, vegetation, spatial, series, scale, zz0307
%X "... examines the effects of using a
segmentation method to identify changepoints or edges in vegetation. It
identifies coherence (spatial or temporal) in place of unconstrained
clustering. ..."
[segmentation]
[doi:10.1016/j.actao.2006.09.002]['07]
Auths 1 & 3 @ Griffith U., auth 2 @ Monash U.
Also see [MML].
%A R. T. O'Donnell
%A L. Allison
%A K. B. Korb
%T Learning hybrid Bayesian networks by MML
%J Advances in Artificial Intelligence
%I SpringerVerlag
%S LNCS/LNAI
%V 4304
%P 192203
%D 2006
%O (19th ACS Australian Joint Conf. on Artificial Intelligence, AI06)
%K conf, c2006, c200x, c20xx, AI06, minimum message, MDL, rodo, description,
LAllison, KBKorb, MML, length, AI, II, classification, CaMML,
Markov Chain Monte Carlo, MCMC, graphical, BDE, graph, network, zz1206,
local structure, CPT, decision tree, trees, Dtree, logit, logistic,
stats, model, models, Bnet, Bnets, net, nets
%X "We use a Markov Chain Monte Carlo (MCMC) MML algorithm to learn hybrid
Bayesian networks from observational data. Hybrid networks represent local
structure, using conditional probability tables (CPT), logit models,
decision trees or hybrid models, i.e., combinations of the three. We compare
this method with alternative local structure learning algorithms using the
MDL and BDe metrics. Results are presented for both real and artificial data
sets. Hybrid models compare favourably to other local structure learners,
allowing simple representations given limited data combined with richer
representations given massive data". isbn:3540497870; isbn13:9783540497875.
 [Bayes nets],
[doi:10.1007/11941439_23]['06],
[pdf]['06].
(Also see [MML].)
%A T. I. Dix
%A D. R. Powell
%A L. Allison
%A S. Jaeger
%A J. Bernal
%A L. Stern
%T Exploring long DNA sequences by information content
%I Probabilistic Modeling and Machine Learning in Structural and
Systems Biology, Workshop Proc.
%W Tuusula, Finland
%P 97102
%M JUN
%D 2006
%K MolBio, wShop, c2006, c200x, c20xx, modelling, models, sequence, MML, MDL,
LAllison, minimum message length, bioinformatics, comparative, chromosome,
description, 1D, TIDix, zz0806, approximate repeats, model, ARM, dotter,
dot plot, plots, 2D
%X ... long ... order of millions of bases, ...
Also see Dix et al, BMC Bioinf., 8(S2):S10, May '07,
[doi:10.1186/147121058S2S10]['08],
and [Bioinformatics].
CR classn '98: F.2, I.2.6, G.3, J.3.; isbn:9521032774.
%A L. Allison
%T A programming paradigm for machine learning with a case study of
Bayesian networks
%J ACSC2006
%P 103111
%M JAN
%D 2006
%K conf, ACSC, ACSW, LAllison, Inductive Programming, IP, c2006, c200x, c20xx,
MML FP, MMLF P, MMLFP, AI, minimum message length, description, MDL, zz0106,
inference, network, net, nets, mixed, generalized, SAR, search and rescue,
SARbayes, general, models, stats, model, trees, II, missing data, Bnet, Bnets
%X "... combines functional programming for writing statistical models &
information theory to prevent overfitting. ... Inductive programming is
illustrated by a case study of Bayesian networks. Networks are built from
classification (decision) trees. Trees are built from partitioning
functions & models on dataspaces. Trees, & hence networks, are general as a
natural consequence of the method. Discrete & continuous variables, &
missing values are handled by the networks. Finally the Bayesian networks
are applied to a challenging data set on lost persons."
 [Bayesian networks],
[Bayesian networks]; also see TR153 c2004,
[paper.pdf] isbn:1920682309.
(Also see [MML].)
%A L. Allison
%T Inductive inference 1.1.2: Inductive programming and a case study of
Bayesian networks
%I Faculty of Info. Tech., Monash University, Clayton, Australia 3800
%R 2005/177
%P 18
%M OCT
%D 2005
%K TR 177, TR177, network, c2005, c200x, c20xx, zz1105, minimum message length,
MML, LAllison, description, machine learning, AI, II, csse, lost person,
FP, MMLFP, MMLF P, missing data, mixed, net, nets, tree, trees, dTree, mdl,
generalized, missing data, TR153, search and rescue, SAR, Bnet, Bnets
%X Builds on the mixed Bayesian nets introduced in TR2004/153.
Also see [Ind.Inf.],
[Ind.Inf.] pages.
%A L. Allison
%T Models for machine learning and data mining in functional programming
%J J. Functional Programming
%I CUP
%V 15
%N 1
%P 1532
%M JAN
%D 2005
%O online at JFP 23/7/2004
%K jrnl, JFP, LAllison, FP, inductive inference, II, AI, c2005, c200x, c20xx,
Haskell, unsupervised, supervised, classification, mixture modelling, stats,
decision tree, dTree, cTree, model trees, minimum message length, MML, type,
zz0105, description, probability, statistical model, MDL, mmld, MMLFP,
MMLF P, Bayesian
%X "... Haskell & its type system are used to define & analyse the
nature of some problems & tools in machine learning & data mining. Data
types & typeclasses for statistical models are developed that allow models
to be manipulated in a precise, typesafe & flexible way. [SMs] considered
inc. prob. distributions, mixture models, functionmodels, timeseries, &
classification & functionmodeltrees. The aim is to improve ways of
designing & programming with models, not only of applying them."
 [doi:10.1017/S0956796804005301][23/7/2004].
Also see [IP],
[II] & [JFP] and
[MML].
%A L. Allison
%T Finding approximate palindromes in strings quickly and simply
%I School of Computer Science and Software Engineering, Monash University,
Australia 3800
%R 2004/162
%M DEC
%D 2004
%K TR 162, TR162, palindrome, MolBio, bioinformatics, algorithm, string,
LAllison, c2004, c200x, c20xx, zz1204
%X Described are two algorithms to find long approximate palindromes in
a string, for example a DNA sequence. A simple algorithm requires
O(n)space and almost always runs in O(k.n)time where n is the
length of the string and k is the number of ``errors'' allowed in the
palindrome. Its worstcase timecomplexity is O(n^2) but this does
not occur with real biological sequences. A more complex algorithm
guarantees O(k.n) worstcase time complexity.
 [arxiv...pdf][12/'04],
[approx.palindromes].
%A D. R. Powell
%A L. Allison
%A T. I. Dix
%T Modelling alignment for nonrandom sequences
%J Advances in Artificial Intelligence
%I SpringerVerlag
%S LNCS/LNAI
%V 3339
%P 203214
%M DEC
%D 2004
%O 17th ACS Australian Joint Conf. on Artificial Intelligence (AI2004)
%K conf, AI, MolBio, dynamic programming algorithms, DPA, c2004, c200x, c20xx,
similar, sequence, homology, Malignment, minimum message length, MML, MDL,
PRSS, FASTA, BLAST, Smith Waterman, context, significance test, P values,
Pvalue, biased, DNA, bias, plasmodium falciparum, low information content,
score, dependent, pattern, repeats, repetitive, shuffling, masking, scoring,
quality, Markov, model, models, hidden, HMM, PHMM, HMMER, description, pair,
strings, DRPowell, LAllison, TIDix, bioinformatics, probabilistic, total,
average
%X "Populations of biased, nonrandom seqs. may cause standard alignment
algorithms to yield falsepositive matches & falsenegative misses. A std
significance test based on the shuffling of sequences is a partial solutions
applicable to pop'ns that can be described by simple models. Maskingout
low information content intervals throws information away. ... new & general
method, modelling alignment: Population models are incorporated into the
alignment process, which can (& should) lead to changes in the rankorder of
matches between a query seq. & a collection of seqs., compared to results
from std algorithms. The new method is general & places very few conditions
on the nature of the models that can be used with it. We apply modelling
alignment to local alignment, global alignment, optimal alignment & the
relatedness problem. Results: As expected, modellingalignment & the
standard PRSS program from the FASTA package have similar accuracy on
sequence populations that can be described by simple models, e.g. 0order
Markov models. However, modellingalignment has higher accuracy on popns that
are mixed or that are described by higherorder models: It gives fewer false
positives & false negatives as show by ROC curves & other results from tests
on real and artificial data". isbn:3540240594.
 [mAlign] inc software,
[doi:10.1007/9783540305491_19]['11].
[Also search for: Allison COMPJ c1999].
%A E. Makalic
%A L. Allison
%A A. Paplinski
%T MML inference of RBF neural networks for regression
%J Brazilian Symp. on Artificial Neural Networks (SBRN)
%I IEEE & Brazillian Computer Soc.
%W Sao Luis
%P 6
%M SEP
%D 2004
%K conf, ANN, network, c2004, c200x, c20xx, Enes, enesm, MML, zz1104, AI, MDL,
minimum message length, radial basis function, RBFs, functions, description,
LAllison, artificial, AIC
%X Also see [MML],
Conf [sbrn] 29/91/10/2004, isbn:8589029042.
%A L. Allison
%T Inductive inference 1.1
%I School of Computer Science and Software Engineering, Monash University,
Australia 3800
%R 2004/153
%P 16
%M MAY
%D 2004
%K TR 153, TR153, TR148, 148, MML, MDL, MMLFP, MMLF, minimum message length,
functional programming, FP, description, CSSE, zz0504, LAllison, AI, hybrid,
Bayesian networks, network, mixture, classification, cluster, clustering,
sequence, classification decision trees, tree, dTree, mixed, model, II,
inductive inference, graphical, statistical model, machine learning, general,
tree, models, net, nets, TR177, 177, c2004, c200x, c20xx, Bnet, Bnets
%X "... Case studies illustrate the [F] style of programming for [II]. ...
Mixtures of Markov Models, Stateful TimeSeries, Bayesian network ..."
See [IP],
[more],
[TR153.pdf].
Also see TR2003/148, TR2005/177 and ACSC2006.
%A L. Allison
%T Inductive inference 1
%I School of Computer Science and Software Engineering, Monash University,
Australia 3800
%R 2003/148
%M DEC
%D 2003
%K TR148, TR 148, MML, MDL, II, MMLFP, MMLF, minimum, message, length, zz1203,
functional programming, FP, data, type, class, description, CSSE, LAllison,
statistical models, Bayesian, artificial intelligence, AI, machine learning
%X See [IP],
[II/200309/].
Also see TR2004/153, JFP c2005, and TR2005/177.
%A E. Makalic
%A L. Allison
%A D. L. Dowe
%T MML inference of singlelayer neural networks
%J Proc. of the 3rd IASTED Int. Conf. on Artificial Intelligence and
Applications
%W Benalmadena, Spain
%P 636642
%M SEP
%D 2003
%O TR 2003/142, CSSE, Monash University, Australia OCT 2003
%K conf, IASTED AIA, ANN, artificial neural network, minimum message length, II,
LAllison, NN, MML, MDL, DLD, Enes, enesm, description, c2003, c200x, c20xx,
Bayesian, Hessian, TR142, TR 142
%X "The architecture selection problem is of great importance when designing
neural networks. A network that is too simple does not learn the problem
sufficiently well. Conversely, a larger than necessary network presumably
indicates overfitting & provides low generalisation performance. This paper
presents a novel architecture selection criterion for single hidden layer
feedforward networks. The optimal network size is determined using a version
of the Minimum Message Length (MML) inference method. Performance is
demonstrated on several problems & compared with a Minimum Description
Length (MDL) based selection criterion." isbn:0889863903; issn:14827913.
[reprint.ps],
[more].
Also see: [TR 2003/142.ps] MML inference of singlelayer neural networks,
and [MML].
%A L. Allison
%T Longest biased interval and longest nonnegative sum interval
%J J. Bioinformatics
%V 19
%N 10
%P 12941295
%M JUL
%D 2003
%K MolBio, jrnl, sequence analysis, DNA, RNA, zz0703, c2003, c200x, c20xx,
bias, density, skew, constraints, low information content, substring,
algorithm, program, locating, maximum, maximal, heaviest, longest, segment,
segments, intervals, regions, positive sum, nonnegative total, AT CG rich,
ATrich, CGrich, problem, CpG, LBI, CABIOS, biased interval, LAllison,
average, score, scores, highest
%X ... an algorithm to find the longest interval having at least a specified
minimum bias in a sequence of characters (bases, amino acids),
e.g. 'at least 0.95 (A+T)rich'. It is based on an algorithm to find
the longest interval having a nonnegative sum in a sequence of positive
and negative numbers. In practice, it runs in linear time; this can be
guaranteed if the bias is rational. ...
 [LBI, inc' code],
[LBI, inc' code].
[can analyse long sequences, e.g. whole chromosomes, very quickly.]
 [doi:10.1093/bioinformatics/btg135]['05],
[.pdf@bioinformatics.oxfordjournals.org][7/'03],
[.pdf@bioinformatics.oxfordjournals.org][7/'03].
%A L. J. Fitzgibbon
%A D. L. Dowe
%A L. Allison
%T Bayesian posterior comprehension via message from Monte Carlo
%J 2nd Hawaii Int. Conf. on Statistics and Related Fields (HICS)
%M JUN
%D 2003
%K conf, HICS, HICS2, HICS2003, statistics, Las Vegas, stochastic, MonteCarlo,
MCMC, posterior probability distribution, LAllison, c2003, c200x, c20xx
%X [Monte Carlo][5/'03],
[reprint.pdf][7/'04].
%A L. Allison
%T The types of models
%J 2nd Hawaii Int. Conf. on Statistics and Related Fields (HICS)
%M JUN
%D 2003
%K HICS, HICS2, c2003, c200x, c20xx, MML, MMLFP, MMLF P, MML F P, FP, MDL,
conf, stats, type, model, II, AI, LAllison
%X [abs],
[.ps],
Also see [Ind. Prog.].
%A L. J. Fitzgibbon
%A L. Allison
%A J. W. Comley
%T Probability model type sufficiency
%J Intelligent Data Engineering and Automated Learning
%I Springer
%S LNCS
%V 2690
%P 530534
%M MAR
%D 2003
%O (4th Int. Conf. on Intelligent Data Engineering and Automated Learning,
IDEAL2003, Hong Kong)
%K conf, IDEAL, IDEAL4, IDEAL2003, statistical model, sufficient statistics,
stats, c2003, c200x, c20xx, LAllison
%X "We investigate the role of sufficient statistics in generalized
probabilistic data mining & machine learning software frameworks. Some issues
involved in the specification of a statistical model type are discussed & we
show that it is beneficial to explicitly include a sufficient statistic &
functions for its manipulation in the model type's specification. Instances
of such types can then be used by generalized learning algorithms while
maintaining optimal learning time complexity. Examples are given for
problems such as incremental learning & data partitioning problems (e.g.,
changepoint problems, decision trees & mixture models)"
[more],
[more][5/'03],
[reprint.pdf][7/'04],
[doi:10.1007/9783540450801_72]['17],
Hong Kong, 2123 March 2003, EUR99.0; uk us isbn:354040550X.
%A J. W. Comley
%A L. Allison
%A L. J. Fitzgibbon
%T Flexible decision trees in a general datamining environment
%J Intelligent Data Engineering and Automated Learning
%I Springer
%S LNCS
%V 2690
%P 761767
%M MAR
%D 2003
%O (4th Int. Conf. on Intelligent Data Engineering and Automated Learning,
IDEAL2003, Hong Kong)
%K conf, IDEAL, IDEAL4, IDEAL2003, decision tree, DTree, c2003, c200x, c20xx,
classification trees, CDMS, LAllison, monash, MML, MDL, AI, II
%X "We describe a new datamining platform, CDMS, aimed at the streamlined
development, comparison and application of machine learning tools. We discuss
its type system, focussing on the treatment of statistical models as
firstclass values. This allows rapid construction of composite models 
complex models built from simpler ones  such as mixture models,
Bayesian networks and decision trees. We illustrate this with a flexible
decision tree tool for CDMS which rather than being limited to discrete
target attributes, can model any kind of data using arbitrary probability
distributions."
 [Dtrees],
[Dtrees],
[reprint.pdf],
[doi:https://doi.org/10.1007/9783540450801_102]['17],
Hong Kong, 2123 March 2003, EUR99.0; uk us isbn:354040550X.
%A L. Allison
%T Types and classes of machine learning and data mining
%J 26th Australasian Computer Science Conference (ACSC)
%P 207215
%M FEB
%D 2003
%O ACS Series Conferences in Research and Practice in
Information Technology V16;
Australian Computer Science Communications, V25, #1, 2003.
%K conf, ACSC, ACSC26, ACSC2003, type, class, check, model, models, regression,
time series, timeSeries, functional programming, FP, MML, MMLF P, MMLFP,
Haskell, type, class, Haskell98, minimum message length, description,
semantics, MDL, formal, inductive inference, stats, statistics, AI, II,
zz0203, MDL, c2003, c200x, c20xx, LAllison
%X [paper],
[.ps].
uk us isbn:0909925941; issn:14451336.
%A L. J. Fitzgibbon
%A D. L. Dowe
%A L. Allison
%T Message from Monte Carlo
%R 2002/107
%I School of Computer Science and Software Engineering, Monash University,
Australia 3800
%M DEC
%D 2002
%K TR 107, TR107, minimum message length, MML, description, MDL, inference, II,
Markov Chain Monte Carlo, MCMC, MMLD, reversible, jump, approximation,
parameter estimation, AI, Bayesian, Montecarlo, Las Vegas, Lasvegas,
model, modelling, c2002, c200x, c20xx, zz1202, LAllison, univariate,
polynomial regression, importance sampling
%X "... [MML] is used to partition a sample from an importance sampling density
of the model parameters into coding regions. The regions contain models that
are similar & hence the method can be loosely described as model clustering.
... The method is made practical by using the MML Boundary Rule & importance
sampling. The MML Boundary Rule is a result that applies to MML
approximations in a common form & which states that the boundary of the
optimal coding region will be isometric or approximately isometric. This
means that we need only consider regions with isometric boundaries.
Importance sampling is used to efficiently evaluate expectations & integrals
using Monte Carlo integration, ... The method requires a sample from a
suitable importance sampling density & a means of approximating the
KullbackLeibler distance between any two models. For general use, we
suggest that the posterior is suitable for the importance sampling. In this
case the Message from Monte Carlo methodology becomes aligned with Bayesian
posterior sampling & can be considered as a means of introspection into the
posterior sample.
We give an example of the method applied to univariate polynomial regression.
[It] was chosen because it is a useful & intuitive problem for which the
MML87 method can & already has been applied. We use the posterior as an
importance sampling density & the polynomial parameters are sampled using
Reversible Jump Markov Chain Monte Carlo."
Also see ICML2002, HICS2003 & [MML];
[abs][12/'02].
%A L. Allison
%T Model classes
%I School of Computer Science and Software Engineering, Monash University,
Australia 3800
%R 2002/125
%M NOV
%D 2002
%K TR 125, TR125, c2002, c200x, c20xx, zz1102, LAllison
%X Also see [ACSC26 '03]
and [II].
%A L. J. Fitzgibbon
%A D. L. Dowe
%A L. Allison
%T Changepoint estimation using new minimum message length approximations
%J Seventh Pacific Rim Int. Conf. on Artificial Intel. (PRICAI2002)
%W Tokyo
%I SpringerVerlag
%S LNAI
%V 2417
%P 244254
%M AUG
%D 2002
%S LNAI
%K conf, PRICAI, PRICAI7, PRICAI02, PRICAI2002, AI, fairly strict, FSMML, MMLA,
minimum message length, MML, inductive inference, II, description, MDL,
segmentation, MMLD, LNAI, LAllison, DLDowe, Monash, csse, zz0902, c2002,
c200x, c20xx
%X Authors at Comp. Sci. and S.W.E., Monash Uni., .au
[abstract]; uk us isbn:3540440380.
[reprint.html],
[@springer][10/'02].
%A L. J. Fitzgibbon
%A D. L. Dowe
%A L. Allison
%T Univariate polynomial inference by Monte Carlo message length approximation
%J Int. Conf. Machine Learning 2002
%W Sydney
%I MorganKaufmann
%P 147154
%M JUL
%D 2002
%K conf, ICML, ICML2002, ICML02, AI, inductive inference, II, LAllison, DLDowe,
Minimum Message Length, MML, description, MDL, Las Vegas, MonteCarlo, MMC,
Markov chain, MCMC, polynomial, fit, fitting, regression, Gibbs sampling,
FSMML, SRM, univariate polynomial, regression, zz0702, c2002, c200x, c20xx
%X Message from Monte Carlo (MMC) algorithm ... compared with
MML87 and structural risk minimization (SRM).
Authors: Comp. Sci. and S.W.E., Monash Uni., .au. Also see TR 2002/107.
[abstract], issn:10491910.
[reprint.html],
[MML].
%A L. Stern
%A L. Allison
%A R. L. Coppel
%A T. I. Dix
%T Discovering patterns in Plasmodium falciparum genomic DNA
%J Molecular and Biochemical Parasitology
%V 118
%N 2
%P 175186
%M DEC
%D 2001
%K MolBio, LAllison, jrnl, c2001, c200x, c20xx, Plasmodium falciparum,
minimum message length, description, Markov, MDL, MBP, MML, Malaria, entropy,
sequence analysis, HMM, PFSA, PFSM, information content, genome, DNA,
pattern, repeat, repeats, DNA models, AED, approximate repeats model, repeat,
pattern discovery, knowledge, hidden, GHMM, Bayesian, LStern, TIDix,
bioinformatics
%X "A method has been developed for discovering patterns in DNA sequences.
Loosely based on the wellknown Lempel Ziv model for text compression, the
model detects repeated seqs. in DNA. The repeats can be forward or inverted,
and they need not be exact. The method is particularly useful for detecting
distantly related seqs., and for finding patterns in seqs. of biased
nucleotide composition, where spurious patterns are often observed because
the bias leads to coincidental nucleotide matches. We show here the utility
of the method by applying it to genomic seqs. of Plasmodium falciparum. A
single scan of chromosomes 2 and 3 of P.falciparum, using our method and no
other a priori information about the seqs., reveals regions of low complexity
in both telomeric and central regions, long repeats in the subtelomeric
regions, and shorter repeat areas in dense coding regions. Application of the
method to a recently sequenced contig of chromosome 10 that has a
particularly biased base composition detects a long internal repeat more
readily than does the conventional dot matrix plot. Space requirements are
linear, so the method can be used on large sequences. The observed repeat
patterns may be related to largescale chromosomal organization and control
of gene expression. The method has general application in detecting patterns
of potential interest in newly sequenced genomic material."
 [paper],
[paper.pdf],
[doi:10.1016/S01666851(01)003887][6/'04].
More on [compression] and
[Bioinformatics].
%A L. J. Fitzgibbon
%A L. Allison
%A D. L. Dowe
%T Minimum message length grouping of ordered data
%J Proc. 11th Int. Workshop on Algorithmic Learning Theory (ALT'2000)
%E H. Arimura
%E S. Jain
%E A. Sharma
%P 5670
%W Sydney
%S LNCS
%V 1968
%M DEC
%D 2000
%K conf, minimum message length, MML, segment, minimum description length, MDL,
ALT9, ALT2000, cut point, points, segmenting, time series, sequence analysis,
complexity, measurement accuracy, model, LAllison, DLDowe, BIC, Monash, II,
segmentation, problem, data, level, levels, c2000, c200x, c20xx
%X Given a series of multivariate data, approximate it by a piecewise
constant function. How many cutpoints are there? What are the means
and variances of each segment? Where should the cut points be placed?
The simplest model is a single segment. The most complex model has
one segment per data point. The best model is generally somewhere
between these extremes. Only by considering model complexity can
a reasonable inference be made.
[more],
[reprint.ps], uk us isbn:3540412379,
[doi:10.1007/3540409920_5]['11].
%A D. R. Powell
%A L. Allison
%A T. I. Dix
%T Fast, optimal alignment of three sequences using linear gap costs
%J J. Theor. Biol.
%V 207
%N 3
%P 325336
%M DEC
%D 2000
%K jrnl, JTB, MolBio, Biology, multiple, sequence, alignments, similarity,
affine, linear, cost, gaps, insert, delete, indel, indels, DNA, time, speed,
fast, string, strings, iterative, phylogenetic, family tree, Ukkonen,
edit distance, dynamic programming algorithm, DPA, DRPowell, LAllison, TIDix,
c2000, c200x, c20xx, zz1100, J Theoretical Biology, bioinformatics
%X [...] The obvious dynamic programming algorithm for optimally
aligning k sequences of length n runs in O(n^k) time. This is
impractical if k >= 3 and n is of any reasonable length.
[...] new algorithm [is] guaranteed to find the optimal alignment [...]
particularly fast when the (threeway) edit distance is small. [...]
O(n + d^3) on average.
[paper][11/'00] and code,
[doi:10.1006/jtbi.2000.2177]['04]
more on [bioinformatics].
%A L. Allison
%A D. Powell
%A T. I. Dix
%T Modelling is more versatile than shuffling
%R 2000/83
%I School of Computer Science and Software Engineering, Monash University,
Australia 3168
%D 2000
%K MolBio, pair, two, sequence, alignment, PFSA, hidden Markov model, PHMM,
low medium information content, repeat, repetition, structure, pattern,
model, shuffle, shuffling, randomize, DNA, permute, tuples, frequencies,
family, PFSM, HMM, dynamic programming algorithm, DPA, homology, algorithm,
minimum message length, MML, description, MDL, LAllison, DRPowell, TIDix,
mAlignment, TR 83, TR83, c2000, c200x, c20xx, bioinformatics
%X It is shown how to incorporate almost any (left to right) model of a
population of sequences into the alignment DPA. Doing so is an alternative
to shuffling/ randomizing (Fitsch; Altschul, Erickson ...) the sequences
to correct for population biases. The resulting algorithm gives
fewer false positives, fewer false negatives, and can (and should)
change the rank ordering of alignments.
[also search for: modelling alignment]
[more],
[mon], and
[Bioinformatics].
%A L. Allison
%A L. Stern
%A T. Edgoose
%A T. I. Dix
%T Sequence complexity for biological sequence analysis
%J Computers and Chemistry
%O (jrnl later became Computational Biology and Chemistry, c2006)
%V 24
%N 1
%P 4355
%D 2000
%K jrnl, comp chem, MolBio, c2000, c200x, c20xx, data compression, compress,
MML, MDL, DNA, mine, approximate repeats, repeat, AED, model, repetitive,
complexity, high, moderate, low, information content, bioinformatics,
LAllison, TIDix, computational biology and chemistry, LStern, description,
minimum message length, bioinformatics, AI, inductive inference,
data mining, time series, timeSeries
%X "A new statistical model for DNA considers a sequence to be a mixture of
regions with little structure and regions that are approximate repeats of
other subsequences, i.e. instances of repeats do not need to match each other
exactly. Both forward and reversecomplementary repeats are allowed. The
model has a small number of parameters which are fitted to the data. In
general there are many explanations for a given sequence and how to compute
the total probability of the data given the model is shown. Computer
algorithms are described for these tasks. The model can be used to compute
the information content of a sequence, either in total or base by base. This
amounts to looking at sequences from a datacompression point of view and it
is argued that this is a good way to tackle intelligent sequence analysis
in general."
 [more],
[paper.pdf],
[doi:10.1016/S00978485(00)800066]['02].
Also see [Bioinformatics].
%A L. Allison
%T Generator and search objects in Java
%J J. Res. and Practice in Inf. Tech.
%V 32
%P 312
%N 1
%D 2000
%K jrnl, ACS, JRPIT, continuation, c2000, c200x, c20xx, LAllison, object, OOP,
generators
%X [.pdf],
more on [Java];
also see TR97/317.
%A T. Edgoose
%A L. Allison
%T MML Markov classification of sequential data
%J Stats. and Comp.
%I Kluwer Academic
%V 9
%N 4
%P 269278
%M SEP
%D 1999
%K jrnl, stats, computing, minimum message length, MML, description, MDL,
statistical, inductive inference, clustering, classification, II, AI,
sequence, data analysis, model, algorithm, mining, series, forwards,
backwards, LAllison, c1999, c199x, c19xx, Snob, time series, timeSeries
%X [MarkovModels],
[paper].
%A D. R. Powell
%A L. Allison
%A T. I. Dix
%T A versatile divide and conquer technique for optimal string alignment
%J IPL
%V 70
%N 3
%P 127139
%D 1999
%K IPL, jrnl, c1999, c199x, c19xx, zz0899, dynamic programming algorithm, DPA,
MolBio, DRPowell, LAllison, TIDix, bioinformatics, space, complexity,
strings, Edit Distance, similarity, LCS, Fast, Hirschberg, Ukkonen, Myers,
time, speed, Linear Space, Check Point, pointing, checkpoint, algorithms
%X A checkpointing (CP) technique uses O(n) space but is simpler
than Hirschberg's O(n)space technique; H' ('97) attributes
an O(N**2)time simple editdistance CP to Eppstein. Here, CP
is applied to more complex cost functions, e.g., linear gap costs, and ...
to Ukkonen's O(n*d)time DPA, even including linear gap costs,
to give O(n)space, O(n.log d + d**2)averagetime,
effectively O(d**2)time in many practical situations.
 [doi:10.1016/S00200190(99)000538][6/'04],
& [DivideandC.].
%A L. Allison
%A D. Powell
%A T. I. Dix
%T Compression and approximate matching
%J COMPJ
%V 42
%N 1
%P 110
%D 1999
%K jrnl, MolBio, COMPJ, Computer Journal, LAllison, DRPowell, TIDix, pair,
string, strings, sequence, alignment, analysis, algorithm, homology,
similarity, limits, HMM, MML, MDL, II, normalized, limit, significance test,
testing, jie, med, DPA, low information content, repeats, repetitive, wei,
nonrandom, nonrandom, compressible, bioinformatics, pair, probabilistic,
information theory, features, complexity, time, fast, speed, shuffling,
shuffle, randomize, DNA, edit distance, hidden Markov model, HMM, PHMM,
c19xx, c1999, c199x, modelling, Malignment, context dependent, scoring
%X A population of sequences is called nonrandom if there is a statistical
model and an associated compression algorithm that allows members of the
population to be compressed, on average. Any available statistical model
of a population should be incorporated into algorithms for alignment of
the sequences and doing so changes the rankorder of possible alignments
in general. The model should also be used in deciding if a resulting
approximate match between two sequences is significant or not. It is
shown how to do this for two plausible interpretations involving pairs
of sequences that might or might not be related. Efficient alignment
algorithms are described for quite general statistical models of sequences.
The new alignment algorithms are more sensitive to what might be termed
`features' of the sequences. A natural significance test is shown to be
rarely fooled by apparent similarities between two sequences that are merely
typical of all or most members of the population, even unrelated members.
 [more],
[doi:10.1093/comjnl/42.1.1]['06],
[pdf@compj]['05].
Also see [Powell AI2004],
and [bioinformatics].
%A L. Allison
%A T. Edgoose
%A T. I. Dix
%T Compression of strings with approximate repeats
%J Intell. Sys. in Mol. Biol. '98
%P 816
%M JUN
%D 1998
%K conf, MolBio, c1998, c199x, c19xx, LAllison, Monash, ISMB6, ISMB98, ISMB 98,
DPA, inductive inference, II, DNA, Chen, Behzadi, F. Le Fessant, protein,
sequence analysis, data, text, repeat, compression, bioinformatics, Li,
repetitive, kwong, hidden Markov models, HMM, low information content,
lines, ALU, sines, ming, self similarity, AED, model, minimum message length,
MML, description, MDL, compress, algorithm
%X "We describe a model for strings of characters that is loosely based on the
Lempel Ziv model with the addition that a repeated substring can be an
approximate match to the original substring; this is close to the situation
of DNA, for example. ... The model has a small number of parameters and these
can be estimated from the given string by an expectationmaximization (EM)
algorithm. Each iteration of the EM algorithm takes O(n^2) time and a few
iterations are typically sufficient. O(n^2) complexity is impractical for
strings of more than a few tens of thousands of characters and a faster
approximation algorithm is also given. The model is further extended to
include approximate reverse complementary repeats when analyzing DNA
strings. Tests include the recovery of parameter estimates from known
sources and applications to real DNA strings."
 [more] ISMB98, 28 June  1 July 1998,
[.pdf] Montreal, Canada; uk us isbn:1577350537.
Also see [bioinformatics].
%A L. Allison
%T Informationtheoretic sequence alignment
%I School of Computer Science and Software Engineering, Monash University
%R 98/14
%M JUN
%D 1998
%K TR14, TR 14, MolBio, LAllison, string, strings, similarity, editdistance,
homology, approximate match, matching, DNA, DPA, hidden Markov model, HMM,
low information content, repeats, repetitive, compressible, MML, MDL,
data compression, content, sequences, c1998, c199x, c19xx, bioinformatics
%X [TR98/14],
[TR98/14].
Also see: Allison, Powell and Dix, `Compression and Approximate Matching',
Comp. J. 42(1) pp110, 1999, for a fuller explanation and later results.
Also see [Bioinformatics].
%A G. Pringle
%A L. Allison
%A D. L. Dowe
%T What is a tall poppy among web pages?
%J Proc. 7th Int. World Wide Web Conference (in Comp. Networks and ISDN Sys.)
%V 30
%N 17
%P 369377
%M APR
%D 1998
%O Comp. Networks and ISDN Systems (Jrnl)
%I Elsevier
%K conf, WWW, WWW7, WWW98, WWWC, WWWC7, WWWC98, IWWWC, IWWWC7, IWWWC98,
decision tree, DT, search engine, engines, rating, ranking, Lycos, Infoseek,
Altavista, Alta vista, page, pages, LAllison, DLDowe, GPringle, Monash,
c1998, c199x, c19xx, zz0598
%X [more],
[more].
Uses inductive inference technology to infer why some web
pages rank highly on certain Internet Search Engines.
Conf. held 1418 April 1998, Brisbane, Australia. uk us isbn:0169755298.
%A D. L. Dowe
%A L. Allison
%A G. Pringle
%T The hunter and the hunted  modelling the relationship between web pages
and search engines
%J 2nd PacificAsia Conf. on Knowledge Discovery and Data Mining (PAKDD98)
%P 380382
%M APR
%D 1998
%I SpringerVerlag
%S LNAI
%V 1394
%K conf, PAKDD2, PAKDD 98, internet, web page, decision tree, engine,
Dtree, LAllison, GPringle, DLDowe, Monash, c1998, c199x, c19xx
%X uk us isbn:3540643834. Also see Pringle et al in WWW7 pp.369377 c1998.
%A T. Edgoose
%A L. Allison
%T Minimum message length hidden Markov modelling
%J Data Compression Conf.
%W Snowbird, Utah
%I IEEE
%P 169178
%M MAR
%D 1998
%K conf, DCC, DCC98, model, MML, MDL, HMM, inductive inference, II, Rissanen,
sequence, model, series, clustering, classification, data analysis, CSci,
description, complexity, time series, timeSeries, CompSci, LAllison, Monash,
Snob, c1998, c199x, c19xx, Snowbird, AI
%X "This paper describes a minimum message length (MML) approach to finding the
most appropriate hidden Markov model (HMM) to describe a given sequence of
observations. An MML estimate for the expected length of a twopart message
stating a specific HMM and the observations given this model is presented
along with an effective search strategy for finding the best number of
states for the model. The information estimate enables two models with
different numbers of states to be fairly compared which is necessary if the
search of this complex model space is to avoid the worst locally optimal
solutions. The general purpose MML classifier 'Snob' has been extended and
the new program 'tSnob' is tested on 'synthetic' data and a large 'real
world' dataset. The MML measure is found to be an improvement on the Bayesian
information criteria (BIG) and the unsupervised search strategy."
 [MarkovM.],
[doi:10.1109/DCC.1998.672145]['18],
[MarkovM.].
%A T. Edgoose
%A L. Allison
%T Unsupervised Markov classification of sequence data using MML
%J Proc. 21st Australian Comp. Sci. Conf.
%E C. McDonald
%W Perth
%I SpringerVerlag
%P 8194
%M FEB
%D 1998
%K ACSC, ACSC21, ACSC98, inductive inference, II, time series, model, HMM,
serial correlation, SNOB, AI, machine learning, hidden, data mining, MDL,
LAllison, Monash, class, cluster, clustering, time series, timeSeries,
c1998, c199x, c19xx, zz0198
%X also see [MML],
uk us isbn:9813083905; order from...[springer].
%A D. R. Powell
%A L. Allison
%A T. I. Dix
%A D. L. Dowe
%T Alignment of low information sequences
%J Australian Computer Science Theory Symposium, CATS '98
%W Perth
%P 215230
%I NUS
%M FEB
%D 1998
%K conf, MolBio, align, HMM, DPA, DNA, ACSC, CATS, CATS98, LAllison, DLDowe,
bioinformatics, probability, low information, simple, features,
TIDix, Monash, c1998, c199x, c19xx, DRPowell
%X "Alignment of two random sequences over a fixed alphabet can be shown to be
optimally done by a Dynamic Programming Algorithm (DPA). It is normally
assumed that the sequences are random and incompressible and that one
sequence is a mutation of the other. However, DNA and many other sequences
are not always random and unstructured, and the issue arises as how to best
align compressible sequences. Assuming our sequences to be nonrandom and
to emanate from mutations of a first order Markov model, we note that
alignment of high information regions is more important than alignment of
low information regions and arrive at a new alignment method for low
information sequences which performs better than the standard DPA for data
generated from mutations of a first order Markov model."
 [more], uk us isbn:9813083921,
[paper.ps]['98].
(Also see [Bioinformatics].)
%A T. Edgoose
%A L. Allison
%A D. L. Dowe
%T An MML classification of protein structure that knows about angles and
sequence
%J Pacific Symposium on Biocomputing '98
%P 585596
%M JAN
%D 1998
%K conf, MolBio, PSB, PSB3, PSB98, von Mises, vonMises, angle, dihedral, class,
cluster, clustering, HMM, SNOB, time series, timeSeries, ARC A49602504,
LAllison, MDL, directional, distribution, bioinformatics, Monash,
c1998, c199x, c19xx
%I World Scientic
%X SNOB + vonMises circular probability distribution + 1st order Markov model.
phipsi pairs give 17 classes and a class seq' correlation matrix.
[paper]
[paper]
[paper.pdf@stanford.edu]['98]; uk us isbn:9810232780.
von Mises, probability density:
f(x  mu, kappa) = (1/(2.pi.I0(kappa))).exp(kappa.cos(xmu))
where I0(kappa) is a normalisation constant.
[Bioinformatics].
%A D. R. Powell
%A D. L. Dowe
%A L. Allison
%A T. I. Dix
%T Discovering simple DNA sequences by compression
%J Pacific Symposium on Biocomputing '98
%P 597608
%M JAN
%D 1998
%K conf, MolBio, PSB, PSB3, PSB98, repeat, repetition, data, model, random,
ARC A49602504, bioinformatics, DRPowell, LAllison, DLDowe, TIDix, Monash,
zz0198, simple, low information, Milosavljevic, Jurka, c1998, c199x, c19xx,
compress
%I World Scientic
%X "An informationtheoretic DNA compression scheme devised by Milosavljevic and
Jurka(1993) has been used in many places in the literature for both the
discovery of new genes and the compression of DNA. Their compression method
applies an encoding of previously occurring runs. They use 5 different
codewords: four being the DNA bases, A,C,G and T, and the other being a
pointer to a previously occurring run. They advocate a codeword of length
log2(5) for each of these and then encoding a run by a codeword of length
2log2(n), where n is the length of the sequence. This scheme encodes the
start of the sequence with a codeword of length log2(n) and likewise encodes
the end of the sequence with a codeword of length log2(n). In this paper,
we show the above coding scheme to be inefficient in various ways and improve
upon it so that it can compress DNA. We discuss our implementation of various
schemes some of which run in linear time."
 [more],
[paper.pdf]['98]
[paper]
uk us isbn:9810232780.
Also see [MML page]
and Milosavljevic '95.
%A K. B. Korb
%A C. Kopp
%A L. Allison
%T Higher education policy in Australia
%J Australian Rationalist
%N 45
%P 1626
%M DEC
%D 1997
%O `A Statement on Higher Education Policy in Australia',
TR 97/318, Department of Computer Science, Monash University, 1997
%K jrnl, TR318, TR 318, HER, West Review, University, Universities, funding,
Government policy, trends, technology, quality, Carlo, LAllison, zz1297,
c1997, c199x, c19xx
%X [.ps.gz],
[html].
%A L. Allison
%T Java continues Prologlike
%I Department of Computer Science, Monash University
%R 97/317
%M APR
%D 1997
%K TR317, TR 317, LAllison, c1997, c199x, c19xx
%X see: Allison, JRPIT, c2000.
%A T. C. Edgoose
%A L. Allison
%T Mixture modelling of sequential datasets
%R 96/285
%I Department of Computer Science, Monash University
%M NOV
%D 1996
%K TR285, TR 285, sequence, series, HMM, hidden Markov model, classification,
clustering, Snob, minimum message length, MML, description, MDL, c1996,
inductive inference, II, data sets, analysis, LAllison, Monash, c199x, c19xx
%X Data is a sequence of items. Each item is a number of measurements.
The program clusters items into classes. Each class is described
by distributions (currently finite state, or Normal, or vonMises)
on attributes. The program determines the number and form of each class.
The class of an item can be influenced by those of its neighbours.
Partial assignment allows for uncertainty of class membership.
Tested on artificial and real data including some weather data
and dihedral angles from protein structures.
[Also search for: Snowbird DCC98],
and see [MML].
%A D. L. Dowe
%A L. Allison
%A T. I. Dix
%A L. Hunter
%A C. S. Wallace
%A T. Edgoose
%T Circular clustering of protein dihedral angles by minimum message length
%J Pacific Symposium on Biocomputing '96
%M JAN
%P 242255
%D 1996
%I World Scientific
%O TR 95/237, Dept. Computer Science, Monash University, Oct 1995
%K PSB, PSB96, TR 237, TR237, Monash, DLD, CSW, CSWallace, LAllison, MolBio,
Monash, classification, angle, von Mises, vonMises, protein structure,
inductive inference, II, MML, MDL, conf, bioinformatics, c1996, c199x, c19xx
%X L. Hunter  NLM, NIH. PSB '96: 36 Jan 1996, Hawaii; uk us isbn:9810225784.
[paper],
[paper.ps][1/'96],
[[eProceedings]][1/'96].
%A R. A. Baxter
%A L. Allison
%A K. Korb
%T Regulation of online information services (a submission)
%M AUG
%D 1995
%K censor, censorship, regulation, control, internet, inter net, Monash,
world wide web, LAllison, www, HREF, http, c1995, c199x, c19xx
%X Submission to the Dept. of Communications and the Arts re their
consultation paper on the regulation of online information services
of 7 July 1995.
Also to the Senate Enquiry into Regulation of Computer OnLine Services
(closing date for submissions 29 Sept 1995)
See also TR224.
%A L. Allison
%T Towards modelling evolution = mutation modulo selection in sequence
alignment
%R 95/225
%I Dept. Computer Science, Monash University
%M JUN
%D 1995
%K LAllison, Monash, TR225, TR 225, MolBio, evolution, pressure, selection,
fit, fitness, family, phylogenetic, evolutionary tree, trees, sequence,
multiple alignment, zz0795, c1995, c199x, c19xx, bioinformatics
%X [bioinformatics].
%A L. Allison
%A R. A. Baxter
%T Protecting Our Innocents
%R 95/224
%I Dept. Computer Science, Monash University
%M JUN
%D 1995
%K LAllison, Monash, TR224, TR 224, internet, inter net, world wide web,
child, children, kids, K12, porn, pornography, violence, politics, sex,
online, censorship, censor, ethics, ethical, CICA, PICS, surfwatch,
www, HREF, http, zz0695, topical, c1995, c199x, c19xx
%X Argues that internet authors (www, ftp, ...) should be encouraged,
NOT required, to describe their works using annotations that are
objective and machine readable. (This is not a censorship proposal and
it cannot be (mis)used as one.)
[TR 95/224 here]
Also see: Regulation of online information services (a submission) Aug 1995.
%A L. Allison
%T Toward a Denotational Semantics for
<A HREF="http://www.cs.monash.edu.au/~lloyd/tildeFP/LambdaLog">LFP</A>
= Logic + Functional Programming, Coded in Lazy ML.
%R 94/204
%M SEP
%D 1994
%I Dept. Computer Science, Monash University
%K LAllison, logic, functional, programming, LP, FP, LFP, FLP, LambdaLog,
LML, Monash, TR 204, TR204, c1994, c199x, c19xx
%X [LFP],
[FP].
%A L. Allison
%T Using Hirschberg's algorithm to generate random alignments of strings
%J IPL
%V 51
%N 5
%P 251255
%M SEP
%D 1994
%K c1994, c199x, c19xx, LAllison, Monash, jrnl, IPL, MolBio, bioinformatics, GS,
DNA, methods, MML, minimum message length encoding, II, inductive inference,
Hirschberg, string, sequence, alignment, similarity, homology, approximate,
match, matching, LCS, LCSS, edit distance, Bayesian, Gibbs sampling, MCMC,
random, sample, DPA, simulated annealing, SA, dynamic programming algorithm,
divide and conquer, hidden Markov model, HMM, probability,
posterior distribution, stochastic
%X Hirschberg's (CACM '75) recursive divide and conquer technique for
the dynamic programming technique (LCS, LCSS, Edit Distance) is
applied to the problem of sampling alignments of two strings
at RANDOM from the alignments' posterior probability distribution.
[more],
[reprint.ps],
[doi:10.1016/00200190(94)900043]['04].
Also see [Bioinformatics].
%A L. Allison
%A C. S. Wallace
%T An information measure for the string to string correction problem with
applications
%J 17th Australian Comp. Sci. Conf.
%P 659668
%M JAN
%D 1994
%W Christchurch, N. Z.
%K LAllison, CSW, CSWallace, Monash, conf, MolBio, inductive inference, II,
string, sequence, family, evolutionary, phylogenetic, tree, trees,
variation, variance, uncertainty, estimate, estimation, parameters, DNA,
multiple alignment, Gibbs sampling, sample, GS, simulated annealing SA,
minimum message length MML, Bayesian, temperature, cooling, probabilistic,
NZ, New Zealand, c1994, c199x, c19xx, ACSC 17, 94, ACSC17, ACSC94,
bioinformatics, Monash
%O Australian Comp. Sci. Comm., Vol 16, No 1(C), 1994, isbn:047302313X.
%X It has been shown how to calculate a probability for an alignment.
Alignments are sampled from their posterior probability distribution.
This is extended to multiple alignments (of several strings). Averaging
over many such alignments gives good estimates of how closely the strings
are related and in what way. In addition, sampling in an increasingly
selective way gives a simulated annealing search for an optimal alignment.
[Bioinformatics],
[paper].
See also the related paper J. Mol. Evol. (39, pp418430, 1994),
"The posterior probability distribution ...", for more results.
%A B. M. Jenkins
%A A. J. Davison
%A L. Allison
%A A. J. Maeder
%T A perimeter based technique for fusion of multisourced micrographic images
%J Dicta93
%W Sydney Australia
%M DEC 810
%D 1993
%P 556563
%K c1993, c199x, c19xx, LAllison, Monash, conf, image processing IP, shape,
curve, signature, DICTA DICTA93, Monash
%X (Also see [MML].)
%A L. Allison
%A C. S. Wallace
%T The posterior probability distribution of alignments and its application
to parameter estimation of evolutionary trees and to optimization of
multiple alignments
%J J. Mol. Evol.
%V 39
%N 4
%P 418430
%M OCT
%D 1994
%O An earlier version is TR 93/188, Dept. Comp. Sci., Monash U., July '93
%K jrnl, MolBio, JME, c1994, c199x, c19xx, LAllison, CSWallace, CSW, DNA,
bioinformatics, optimisation, estimate, infer, parameters, algorithm,
multiple, alignment, data, string, molecular, sequence, homology, Markov,
family, phylogenetic, tree, trees, edit distance, Monte Carlo method, mcmc,
simulated annealing, SA, inductive inference, II, sample, speed, Bayesian,
dynamic programming algorithm, DPA, stochastic, methods, GS, Gibbs sampling,
minimum message length encoding, MML, chain, minimum description length, MDL,
transthyretin, chloramphenicol resistance gene, CAT, CATB, CATD, CATP, CATQ,
CCOLI, ECOLI, algorithmic, mutual information, theory, significance,
probabilistic, temperature, limits, TR 93/188, TR188
%X "It is shown how to sample alignments from their posterior probability
distribution given two strings. This is extended to sampling alignments of
more than two strings. The result is firstly applied to the estimation of
the edges of a given evolutionary tree over several strings. Secondly,
when used in conjunction with simulated annealing, it gives a stochastic
search method for an optimal multiple alignment."
 [paper] and source code,
[reprint.ps],
[doi:/10.1007/BF00160274]['07].
(The JME paper is a much expanded and changed version of TR 93/188,
[TR93/188](.ps))
%A A. Finlay
%A L. Allison
%T A correction to the denotational semantics for Prolog of Nicholson and Foo
%I ACM
%J TOPLAS
%V 15
%N 1
%P 206208
%M JAN
%D 1993
%K TOPLAS, jrnl, note, LAllison, Alan, alanf, Monash, DS, semantic, Prolog,
logic programming, LP, c1993, c199x, c19xx, Monash
%X [doi:10.1145/151646.151652][2/'04]
Also see [Logic Programming],
and [Denotational Semantics].
%A L. Allison
%T Applications of recursively defined data structures
%J Aust. Comp. J.
%V 25
%N 1
%P 1420
%M FEB
%D 1993
%K jrnl, LAllison, Monash, ACJ, functional applicative programming, FP,
lazy, need, circular, program, recursive, recursion, data structure,
list, lists, search trees, tree, breadth first traversal, queue, memo,
Fib, Fibonacci, nqueens, nqueens, n queens, irreducible, good,
sequences, Axel Thue, c1993, c199x, c19xx, TR 88 119 TR88119 TR119
%O an early version in TR 88/119 Dept. Computer Science, Monash University '88
%X A circular program contains a data structure whose definition is
selfreferential or recursive. The use of such a definition allows
efficient functional programs to be written and can avoid the creation of
intermediate data structures that would have to be garbage collected.
This paper uses circular programs in various ways, to implement
memostructures and explicit searchtrees to hold solutions to
constraintsatisfaction problems.
Examples: search trees, nqueens, irreducible sequences.
 [more],
[reprint.ps].
%A L. Allison
%T A fast algorithm for the optimal alignment of three strings
%J J. Theor. Biol.
%V 164
%N 2
%P 261269
%M SEP
%D 1993
%O TR 92/168 Dept. Computer Science, Monash University, Oct '92.
%K LAllison, Monash, jrnl, II, JTB, MolBio, bioinformatics, multiple alignment,
edit distance, Ukkonen, three, string, strings, sequence, sequences,
dynamic programming algorithm, DPA, TR 92 168 TR92168 TR168,
c1993, c199x, c19xx, J Theoretical Biology
%X Given 3 strings, length ~ n, 3way editdistance d,
O(n.d^2) time algorithm worst case, O(n+d^3) typically.
Tree costs 0/1/2. ie. xxx :0; xxy, xx, x :1; xyz, xy :2
NB. Each internal node of an unrooted binary tree has 3 neighbours.
[more],
[reprint.ps],
[paper] inc' pdf paper and code,
[doi:10.1006/jtbi.1993.1153]['04],
and more on [bioinformatics].
%A L. Allison
%T Normalisation of affine gap costs used in optimal sequence alignment
%J J. Theor. Biol.
%M MAR
%D 1993
%V 161
%N 2
%P 263269
%K LAllison, Monash, jrnl, MolBio, JTB, alignment, string, sequence analysis,
edit distance, homology, gap, gaps, linear, indel, insert delete, Altschul,
mutual information, similarity, hidden Markov model, HMM, bioinformatics,
inductive inference, II, c1993, c199x, c19xx, DNA, J Theoretical Biology
%X "It is shown how to normalize the costs of an alignment algorithm that
employs affine or linear gap costs. The normalized costs are interpreted as
the log probabilities of the instructions of a finitestate editmachine.
This gives an explicit model relating sequences that can be linked to
processes of mutation and evolution."
 [more],
[reprint.ps],
[paper] inc' pdf.
%A L. Allison
%T Lazy dynamic programming can be eager
%J IPL
%V 43
%N 4
%P 207212
%M SEP
%D 1992
%K LAllison, Monash, jrnl, FP, lazy functional programming, Haskell, fast,
efficient, dynamic programming algorithm, DPA, edit distance, LCS, LCSS,
MolBio, approximate, similar, string, strings, match, matching, sequence,
alignment, c1992, c199x, c19xx, bioinformatics
%X Lazy evaluation in a functional language is exploited to make the simple
dynamic programming algorithm for the editdistance problem run quickly
on similar strings: being lazy can be fast.
Runs in O(n*d) time thanks to laziness.
 [more],
[reprint.ps]
[html]
[doi:10.1016/00200190(92)902027]['07].
%A C. N. Yee
%A L. Allison
%T Fast string alignment with linear indel costs
%R TR 92/165
%I Computer Science, Monash University
%M JUL
%D 1992
%K LAllison, MolBio, string alignment, similarity, homology, Ukkonen,
edit distance, linear indel gap cost, costs, Ukkonen, Monash,
TR 165, TR92/165, TR165, II, c1992, c199x, c19xx, bioinformatics
%X two strings, O(n*d) time.
The constants in the cost function have to be "small" integers.
[Computing for Molecular Biology].
[Also search for: Ukkonen].
%A D. L. Dowe
%A J. Oliver
%A L. Allison
%A T. I. Dix
%A C. S. Wallace
%T Learning rules for protein secondary structure prediction
%J Proc. 1992 Department Research Conf.
%I Dept. Computer Science, University of Western Australia
%E C. McDonald
%E J. Rohl
%E R. Owens
%M JUL
%D 1992
%O TR 92/163, Dept. Computer Science, Monash University, JUN '92
%K LAllison, CSW, DLD, Monash, UWA, WA, conf, MolBio, decision tree, trees,
graph, protein, amino acid, AA, secondary structure, SS, prediction,
rule, rules, alpha helix, beta strand, extended sheet, coil, turn,
CSWallace, inductive inference, II, MML, minimum message length, c1992,
c199x, c19xx, bioinformatics, TR 92 163, TR92163, TR163
%X [TR92/163.ps]
Also see [Bioinformatics],
and TR 92/163.
[CSci UWA home]['00]; uk us isbn:0864221959.
%A D. L. Dowe
%A J. Oliver
%A T. I. Dix
%A L. Allison
%A C. S. Wallace
%T A decision graph explanation of protein secondary structure prediction
%J 26th Hawaii Int. Conf. Sys. Sci.
%V 1
%P 669678
%M JAN
%D 1993
%K LAllison, CSW, Monash, conf, MolBio, protein secondary structure prediction,
conformation, alpha helix, ss, AA, beta sheet extended strand, turn, coil,
II, inductive inference, decision graph tree, DTree, CSWallace, CSW,
MML, Minimum message length encoding, description, MDL, Bayesian,
TR163 163, c1993, c199x, c19xx, bioinformatics, HICSS, HICSS26, HICSS93
%X Oliver and Wallace (IJCAI '91) introduced `decision graphs' 
a generalisation of decision trees  here applied to protein secondary
structure prediction.
[more],
[paper (HTML)].
Also see TR 92/163.
%A C. N. Yee
%A L. Allison
%T Reconstruction of strings past
%J Bioinformatics
%O (was Comp. Appl. BioSci.)
%V 9
%N 1
%P 17
%M FEB
%D 1993
%O TR 92/162, Dept. Computer Science, Monash University, May '92
%K LAllison, Monash, jrnl, MML, minimum message length, mdl, J. Bioinformatics,
encoding, hidden Markov model, HMM, II, inductive inference, probabilistic,
MolBio, string, sequence, alignment, homology, similarity, edit,
evolutionary, distance, parameter estimation, rtheory,
TR TR92162 TR162 162, c1993, c199x, c19xx, CABIOS, J. Bioinformatics
%X Use of single optimal alignment gives biased estimates of the evolutionary
"distance" between two strings but the rtheory, average all alignments,
recovers accurate estimates over a very wide range of similarity.
 [more],
[doi:10.1093/bioinformatics/9.1.1]['11],
[reprint.ps]
[paper.html].
Also see [Bioinformatics].
[now J. Bioinformatics].
%A L. Allison
%T Some algorithmic attacks on multiple alignment (abstract)
%J Boden Conf.
%W Thredbo, Australia
%M FEB
%D 1993
%K LAllison, Monash, RSBS, ANU, conf, MolBio, MML, II, string,
sequence, approximate match, matching, three, DPA, bioinformatics
%X also see [Bioinformatics].
%A L. Allison
%T Estimating parameters and evolutionary distances in the inference of
evolutionary trees (abstract)
%J Robertson Symposium.
%W Australian National University
%M JAN
%D 1993
%K LAllison, Monash, RSBS, ANU, conf, MolBio, MML, inductive inference,
II, string, sequence, multiple alignment, approximate match,
matching, phylogenetic, tree, trees, c1993, c199x, c19xx, bioinformatics
%X also see [Bioinformatics].
%A L. Allison
%A C. S. Wallace
%A C. N. Yee
%T Minimum message length encoding, evolutionary trees and multiple alignment
%J 25th Hawaii Int. Conf. on Sys. Sci.
%K LAllison, CSW, Monash, conf, MolBio, minimum message length encoding, MML,
ML, evolutionary, family, phylogenetic, tree, trees, CSWallace, CSW, human,
Bayesian, finite state, model, machine, FSM, hidden Markov model, primate,
HMM, DNA, multiple alignment, inductive inference, II, bioinformatics, chimp,
HICSS, HICSS25, HICCS92, TR 91 155, TR91155, TR155, c1992, c199x, c19xx
%V 1
%P 663674
%M JAN
%D 1992
%O TR 91/155, Dept. Computer Science, Monash University '91
%X "A method of Bayesian inference known as MML encoding is applied to inference
of an evolutionary tree and to multiple alignment for K >= 2 strings.
It allows the posterior oddsratio of two competing hypotheses, for
example two trees, to be calculated. A tree that is a good hypothesis forms
the basis of a short message describing the strings. The mutation
process is modelled by finitestate machine. It is seen that tree
inference and multiple alignment are intimately connected."
 [paper],
there is an example on the primate globin pseudogenes.
(Also see [Bioinformatics].)
%A L. Allison
%A T. I. Dix
%A C. N. Yee
%T Shortest path and closure algorithms for banded matrices
%J IPL
%V 40
%P 317322
%M DEC
%D 1991
%O TR 89/133, Computer Science, Monash University, Nov 1989
%K LAllison, Monash, jrnl, all pairs, shortest path, paths, closure, algorithm,
band, banded, matrix, matrices, graph, ve negative cycle problem,
constraint, TR89/133, TR 89, 133, TR133, c1991, c199x, c19xx
%X All pairs shortest paths problem in banded (width b) matrices.
O(n b^2) for entries in band and for negative cycle problem,
O(n^2 b) for all pairs shortest paths where b is band width.
[more],
[HTML]
[.ps]
[doi:10.1016/00200190(91)902002]['07].
%A S. T. S. Ho
%A L. Allison
%A C. N. Yee
%A T. I. Dix
%T Constraint checking for restriction site mapping
%J 24th Hawaii Int. Conf. on Sys. Sci.
%V ?
%P ??????
%M JAN
%D 1991
%O TR 89/129 Computer Science, Monash University, JUL '89
%K LAllison, Monash, conf, MolBio, HICSS, HICSS24, HICSS91, bioinformatics,
restriction site mapping, RSM, map, constraint satisfaction problem CSP,
separation theory, TR 89/129 89 129 TR89/129 TR129, c1991, c199x, c19xx
%X also see [mapping].
%A L. Allison
%A M. R. Garrett
%A A. J. Maeder
%T Particle shape characterization and comparison using curve signatures
%J 21st Conf. of the Fine Particle Society
%M AUG
%D 1990
%C San Diego
%K LAllison, Monash, conf, time warp, warping, shape, particle, closed, curve,
signature, alignment, MML, edit distance, FPS, image processing, IP, c1990
c199x, c19xx
%X (Also see [MML].)
%A S. Ho
%A L. Allison
%A C. N. Yee
%T Restriction site mapping for three or more enzymes
%J Comp. Appl. BioSci. (J. Bioinformatics)
%V 6
%N 3
%P 195204
%M JUL
%D 1990
%O TR 88/117 Dept. Comp. Sci., Monash University Oct '88
%K LAllison, Monash, jrnl, MolBio, CABIOS, RSM, restriction site map, mapping,
constraint satisfaction problem, programming, three, separation theory,
TR 88 117 TR117 TR88/117, c1990, c199x, c19xx, J. Bioinformatics, enzyme
%X "Restriction site mapping requires a generator to put forward possible maps &
a constraint checker to reject false maps. Ideally these combine to give an
algorithm which calculates a sound & complete solution set. 3 algorithms for
generation are presented & compared. Two decompose a multienzyme problem
(3+) into subproblems. The constraint checker is based on separation theory.
Some insights into the extent of constraint checking involved in & the
feasibility of more checking for three or more enzymes are discussed. The
tradeoff between comp'n time & the soundness of the soln set is examined."
[now J. Bioinformatics]
 [doi:10.1093/bioinformatics/6.3.195],
[jrnl]['07].
Also see [mapping].
%A L. Allison
%A Du Xiaofeng
%T Relating three strings by minimum message length encoding (abstract)
%P 13
%J International Conference on Genes, Proteins and Computers
%W Chester
%I SERC Daresbury Laboratory
%M APR
%D 1990
%K LAllison, Monash, conf, MolBio, multiple, three, triple, alignment,
LCS, LCSS, MML, family, evolutionary phylogenetic tree, bioinformatics,
inductive inference, II, DNA, c1990, c199x, c19xx
%X also see [Bioinformatics].
%A L. Allison
%A C. S. Wallace
%A C. N. Yee
%T Finitestate models in the alignment of macromolecules
%J J. Mol. Evol.
%V 35
%N 1
%P 7789
%M JUL
%D 1992
%K LAllison, jrnl, MolBio, c1992, c199x, c19xx, TR 90/148, macromolecules,
TR90/148, TR148, 148 inductive inference, II, DNA, bioinformatics, DPA,
dynamic programming algorithm, mutual information, ML, string, sequence,
comparison, alignment, minimum message length encoding, MML, FSM, FSA,
finite state model, analysis, minimum description length, MDL, methods,
Hidden Markov model, HMM, homology, similarity, LCS, LCSS, significance,
evolutionary, edit distance, sequence, rtheory, linear, gap, indel, insert,
delete, Algorithm, Time, Speed, JME, AAAI, Bayes, Bayesian, CSWallace, CSW
%O An extended abstract titled:
Inductive inference over macromolecules
in joint sessions at AAAI Symposium, Stanford MAR 1990
on (i) Artificial Intelligence and Molecular Biology, p59,
& (ii) Theory and Application of MinimalLength Encoding, p5054,
also an early version in Technical Report 90/148,
Dept. Comp. Sci., Monash U., Australia 3168.
%X MML encoding is a technique of inductive inference with theoretical and
practical advantages. It allows the posterior oddsratio of two theories
or hypotheses to be calculated. Here it is applied to the problem of
aligning or relating two strings, in particular biological macromolecules.
We compare the rtheory, that the strings are related, with the nulltheory,
that they are not related. If they are related the probabilities of the
various alignments can be calculated. This is done for the one, three
and fivestate models of relation or mutation. These correspond to linear
and piecewise linear cost functions on runs of indels. We describe how
to estimate the parameters of a model. The validity of a model is itself
a hypothesis and can be tested objectively. This is done on real DNA
and on artificial data. The tests on artificial data indicate limits on
what can be inferred in various situations. The tests on real DNA support
either the three or the fivestate models over the onestate model.
Finally, a fast, approximate minimum message length string comparison
algorithm is described.
 [doi:10.1007/BF00160262]['07].
[reprint] and software,
See C. S. Wallace & D.M Boulton
An information measure for classification.
CompJ 11(2) 185194 Aug '68 (appendix)
for the derivation of the coding scheme for multistate data.
See also (i) Bishop & Thompson
(ii) Thorne, Kishino & Felsenstein, and
[AIMB](.ps),
[Alignment].
%A L. Allison
%A C. S. Wallace
%A C. N. Yee
%T When is a string like a string?
%J Int. Symposium on Artificial Intelligence and Mathematics
%W Ft. Lauderdale, Florida, USA
%M JAN
%D 1990
%K LAllison, CSW, CSWallace, Monash, conf, inductive inference, II, homology,
alignment, LCS, edit distance, string, sequence, comparison, similarity,
rtheory, macromolecule, MolBio, DNA, uncertainty, pattern matching, MML,
minimum message length encoding, AIM AIM90, Hidden Markov model, HMM,
c1990, c199x, c19xx, bioinformatics
%X [more],
[html],
[.ps](.ps)
also see [TR90/148](html)
and [TR90/148](.ps).
%A L. Allison
%A C. N. Yee
%T Minimum message length encoding and the comparison of macromolecules
%J Bulletin of Mathematical Biology
%V 52
%N 3
%M MAY
%D 1990
%P 431453
%O TR 89/126 Computer Science, Monash University, MAY '89
%K LAllison, Monash, jrnl, minimum message length encoding, MML, c1990, c199x,
c19xx, MDL, ML, inductive inference, II, minimum description length, DNA,
approximate, alignment, similarity, homology, LCS, LCSS, pattern matching,
string, sequence, comparison, Bayesian, Hidden Markov model, HMM,
mutual information, MolBio, bioinformatics, BMB, TR 89/126, 89 126, TR89/126
%X "A method of inductive inference known as minimum message length encoding is
applied to string comparison in molecular biology. The question of whether or
not two strings are related and, if so, of how they are related and the
problem of finding a good theory of string mutation are treated as inductive
inference problems. The method allows the posterior oddsratio of two string
alignments or of two models of string mutation to be computed. The
connection between models of mutation and existing string alignment
algorithms is made explicit. A fast minimum message length alignment
algorithm is also described."
[more],
[doi:10.1007/BF02458580]['06].
%A C. McDonald
%A L. Allison
%T Denotational semantics of a command interpreter and their implementation
in standard ML
%J COMPJ
%V 32
%N 5
%P 422431
%M DEC
%D 1989
%O TR 88/7, Dept. Computer Science, UWA '88
%K LAllison, Monash, COMPJ, jrnl, functional programming, FP, c1989, c198x,
c19xx, shell script, scripting, command language, line, unix,
operating system, OS, denotational semantics, DS, ML, SML, UWA,
TR 88/7, TR88/7, 88 7
%X [doi:10.1093/comjnl/32.5.422]['11],
[paper],
[Denotational Semantics].
%A L. Allison
%T Continuations implement generators and streams
%J COMPJ
%V 33
%N 5
%P 460465
%M OCT
%D 1990
%O TR 88/112 Dept. Computer Science, Monash University AUG '88
%K LAllison, Monash, jrnl, continuation, combinator, generator, nondeterminism,
nondeterminism, nqueens, n queens, nqueens, source, stream, sink, lazy,
agent, sieve, TR 88/112, TR88/112, TR112, c1988, c198x, c19xx,
functional programming, FP
%X Continuations are used to program nondeterministic generators
and a variation on deterministic stream functions.
Examples include the nqueens problem and the sieve of Eratosthenese.
The continuation style has an imperative flavour but is
functional and free of side effects.
Some (recursive) programs can be executed in constant space.
 [doi:10.1093/comjnl/33.5.460]['11],
[paper].
%A L. Allison
%A C. N. Yee
%A M. McGaughey
%T Threedimensional queens problems
%R TR 89/130
%I Dept. Computer Science, Monash University, Australia
%M AUG
%D 1989
%K LAllison, Monash, n queens, nqueens, queen, nqueens, chess, 3D,
TR 89 130 TR130 TR89/130, c1989, c198x, c19xx
%X [TR89/130],
[TR89/130 (HTML)],
[TR89/130 (.ps)].
%A L. Allison
%A C. Y. Yee
%T Restriction site mapping is in separation theory
%J Comp. Appl. BioSci. (J. Bioinformatics)
%V 4
%N 1
%P 97101
%M JAN
%D 1988
%K LAllison, Monash, jrnl, CABIOS, plasmid, map, maps, mapping, combinatorial,
enzyme, restriction site, DNA, phage, RSM, MolBio, separation theory,
linear, constraint, programming, algorithm, inequality, c1988, c198x, c19xx,
J. Bioinformatics, NAR, Nucl. Acids Res. special issue
%X Paper examines constraint checking during backtracking generation
of solutions to the plasmid mapping problem.
The constraints fall into Vaughn Pratt's separation theory.
It is necessary and sufficient to check all simple cycles
in the graph of restriction sites linked by fragments
from the two single digests and the one double digest (can be generalized
to more than two enzymes).
In general, at level 'n' there are at most n new constraints to check.
As an efficiency matter, only those cycles containing 1, 2 or 3
indivisible cycles can be checked which allows a very few false maps to
be generated. This reduces total constraints/map from n*n to 3*n.
 [more],
[doi:10.1093/comjnl/33.5.460]['11],
[paper (HTML)],
[paper (.ps)].
[jrnl is now J. Bioinformatics].
%A L. Allison
%T A Practical Introduction to Denotational Semantics
%S Cambridge Computer Science Texts
%I CUP
%K LAllison, UWA, text, book, textbook, CUP, denotational semantics, DS,
programming language, languages, definition, L4REF, c1986, c198x, c19xx
%D 1986
%X [doi:10.1017/CBO9781139171892];
pb, uk us isbn:0521314232, uk us isbn13:9780521314237;
hb, uk us isbn:0521306892, uk us isbn13:9780521306898;
reviewed in Computing Reviews, Nov '87 pp.5745 and Jul '88 pp.349350
and in Software P & E 18(5) pp.493494 May '88 [www]['07].
(More on [Denotational Semantics].)
%A L. Allison
%T Direct semantics and exceptions define jumps and coroutines
%J IPL
%V 31
%N 6
%P 327330
%M JUN
%D 1989
%O TR 87/90 Dept Computer Science, Monash University, 1987
%K LAllison, Monash, jrnl, denotational semantics, DS, direct, exception,
SML, ML, jump, sequencer, goto, coroutine, control, raise, handle, catch,
throw, IPL, TR 87/90 TR87/90 90 TR90, c1989, c198x, c19xx
%X Direct semantics and continuation semantics are the two main styles of
denotational semantics. Direct semantics is the simpler style but cannot
define the semantics of jumps and other sequencers. This paper shows that,
with the addition of exceptions, direct semantics can define sequencers. In
contrast to the use of continuation semantics, nothing need be added to the
semantics of commands unconnected with sequencers. Since exceptions can be
defined in terms of the lambdacalculus nothing need be added to the
foundations of semantics.
See [TR87/90][HTML],
[doi:10.1016/00200190(89)900975]['06],
[Denotational Semantics].
%A L. Allison
%T Circular programs and selfreferential structures
%J SPE
%V 19
%N 2
%P 99109
%M FEB
%D 1989
%O in TR87/91 Dept Computer Science, Monash University, Jan 87
%K jrnl, LAllison, Monash, applicative, functional programming, FP, circular,
recursive, recursion, lazy, evaluation, need, program, list, lists, queue,
doubly linked, tree, trees, breadth first traversal, BFT, threaded, queues,
nub, unique elements, primes, c1989, c198x, c19xx, TR 87/91 TR87/91 TR91 91
%X A circular program creates a data structure whose computation depends on
itself or refers to itself. The technique is used to implement the
classic data structures circular and doublelinked lists, threaded trees
and queues in a functional programming language.
 [more],
[doi:10.1002/spe.4380190202]['11],
[reprint.ps],
[html].
%A L. Allison
%T Some applications of continuations
%J COMPJ
%V 31
%N 1
%P 911
%M FEB
%D 1988
%O in TR87/91 Dept Computer Science, Monash University, Jan 87
%K LAllison, Monash, jrnl, functional programming, FP, recursion, backtrack,
COMPJ, backtracking, back track, coroutine, applicative, continuation, parse,
parser, parsing, recursive descent, nondeterminism, binary search tree,
merge, combinators, TR 87/91 TR87/91 TR91 91, c1988, c198x, c19xx
%X [doi:10.1093/comjnl/31.1.9]['11],
[paper].
%A L. Allison
%A T. I. Dix
%T A bitstring longestcommonsubsequence algorithm
%J IPL
%V 23
%M DEC
%D 1986
%P 305310
%K LAllison, TIDix, Monash, UWA, jrnl, MolBio, LCS, LCSS, c1986, c198x, c19xx,
fast, algorithm, bits string, bit vector, DPA, dynamic programming algorithm,
similarity, homology, bioinformatics, IPL, sequence, approximate, match,
matching, strings, distance, comparison, alignment, practical, speedup,
speed
%X "A longestcommonsubsequence algorithm is described which operates in terms
of bit or bitstring operations. It offers a speedup of the order of the
wordlength on a conventional computer."
Speedup is ~ wordlength (eg. a factor of 32 or 64), time is O(n^2/wordlength).
[more]+source code,
[HTML],
[.ps],
[doi:10.1016/00200190(86)900918]['07].
%A A. Finlay
%A L. Allison
%T A prescription for Prolog control features based upon denotational semantics
and its implementation in standard ML
%R TR 87/97
%I Dept. Computer Science, Monash University
%M NOV
%D 1987
%K LAllison, Monash, CSSE, Prolog, logic programming, LP, SML, ML, TR87/97,
TR97 TR 97, c1987, c198x, c19xx, denotational semantics, DS
%X [Logic Programming]
[Denotational Semantics].
%A L. Allison
%T Programming denotational semantics II
%J COMPJ
%V 28
%N 5
%P 480486
%D 1985
%K COMPJ, jrnl, LAllison, UWA, denotational, DS, continuation semantics,
algol, A68, algol68, algol68, c1985, c198x, c19xx
%X [paper]
[Denotational Semantics].
%A L. Allison
%T Programming denotational semantics
%J COMPJ
%V 26
%N 2
%P 164174
%D 1983
%K COMPJ, jrnl, LAllison, UWA, direct, denotational semantics,
DS, Pascal, c1983, c198x, c19xx
%X [paper],
[Denotational Semantics].
%A L. Allison
%T An executable Prolog semantics
%J Algol Bulletin
%N 50
%P 1018
%M DEC
%D 1983
%K LAllison, UWA, jrnl, denotational semantics, DS, algol, algol68, algol68,
logic programming, LP, unification, interpreter, Prolog, c1983, c198x, c19xx
%X [doi:10.1145/1061790.1061794]['06]  AB in ACM digital archive.
Also see [program],
and [prolog].
%A L. Allison
%T On syntax directed program editing
%J SPE
%V 13
%P 453465
%D 1983
%K LAllison, UWA, jrnl, edit, editor, syntaxeditor, SED, SDE,
syntax directed, program, structure based, c1983, c198x, c19xx
%X .
%A L. Allison
%T Stablemarriages by coroutines
%J IPL
%V 16
%P 6165
%M FEB
%D 1983
%K IPL, jrnl, c1983, c198x, c19xx, LAllison, UWA, jrnl, stable marriage problem,
algorithm, match, coroutine, modula, two, modula2, modulaII, bipartite,
graph, matching, coroutine
%X The stable marriage problem is an appealing version of many pairing problems.
A solution by coroutines is given, based on the recursive algorithm of
McVitie and Wilson (1971). There are few published algorithms where
coroutines are really useful but they solve this problem very naturally.
 [doi:10.1016/00200190(83)90025X]['06],
[LA home].
%A L. Allison
%A P. Edmiston
%T A LOGO survey
%J Australian Computer Bulletin
%V 5
%N 9
%P 40
%M OCT
%D 1981
%K LAllison, UWA, jrnl, logo, programming language,
children school education CAI, c1981, c198x, c19xx
%A L. Allison
%T Generating coset representatives for permutation groups
%J J. of Algorithms
%V 2
%P 227244
%M SEP
%D 1981
%K LAllison, UWA, jrnl, coset, group, permutation, backtrack, backtracking,
back track, isomorph rejection, subgroup, generate, representative,
algorithm, c1981, c198x, c19xx
%X .
 [doi:10.1016/01966774(81)900249]['10].
%A L. Allison
%T Phrase structures, nondeterminism and backtracking
%J IPL
%V 7
%N 3
%P 139143
%M APR
%D 1978
%K LAllison, MelbU, jrnl, grammar, parse, parsing, parser, backtrack,
backtracking, nondeterminism, back track, back tracking,
LL, LL1, LL(1), c1978, c197x, c19xx
%X  [doi:10.1016/00200190(78)900777]['06],
[ProgrammingLanguages].
%A L. Allison
%A A. Lock
%T A wordprocessor for the handicapped
%J ACSC 5
%W Perth
%M FEB
%D 1982
%K LAllison, UWA, conf, ACSC5, ACSC82, c1982, c198x, c19xx
%A L. Allison
%A D. Wilde
%T Phrase structures in Pascal
%J Programming Language Systems
%W Canberra
%P 2937
%M FEB
%D 1977
%I ANUpress
%K LAllison, MelbU, conf, grammar, parse, parsing, parser, backtrack,
backtracking, back track, c1977, c197x, c19xx, Pascal, LL LL1 LL(1), ACSC,
ACSC0, structure, system
%X Essentially ACSC0, conf 2425 Feb 1977,
volume published in '78, isbn:0708104932.
Also see [ProgrammingLanguages].

window on the wide world:

