<1998<
^CSE2304^
>progress 1999>
CSE423 Plan 1999
NO OFFICIAL STATUS, PLANNING ONLY
DLD's plan B (14/7/1999) and B' (16/7/1999):
The numbering does not necessarily reflect the lecture number.
- probability, discrete, continuous, cumulative, joint, independent,
dependent, marginal,
1_lecture
- binomial distribution
- multinomial distribution
- Normal (Gaussian) distribution 2nd_lecture
- information, entropy
- Kullback-Leibler distance, footy tipping
- (prefix-) codes, examples, Huffman, arithmetic coding 4th_lecture
- integers (codes for), geometric, log*, tree-code, other (Poisson) ...
- inference, model class ... model ... parameter(s), estimation, bias,
invariance
- Bayes, (max-) likelihood, prior, posterior,
- Fisher information matrix, general form of MML
lect_10 and 11 loose on
binomial distribution,
multinomial distribution,
Normal (Gaussian) distribution
- unsupervised classification, Snob.
Use multinomial for class assignments,
total assignment (and inconsistency),
partial assignment and coding trick
- supervised classification, (decision-) trees and graphs,
encode binomial/multinomial tree structure,
encode continuous-valued cut points in tree structure,
encode attribute being split upon,
encode leaf nodes (1_1/2_lectures),
decision graphs (1/2_lecture),
applications - bushfire, proteins (1/2_lecture)
- sequential data, low-order Markov models, Lempel-Ziv, PFSA etc.
- applications: segmentation, ...
- (piece-wise) straight line fitting, linear regression, polygon fitting
- sequences and approximate matching
May or may not get to:
- ?image compression, Peter T?
- ?causal models, Kevin K?
LA's plan A (6/7/1999):
The numbering is Lloyd's cut #1;
it does not reflect lecture number,
nor duration, nor ordering, yet.
Looks like more than I had anticipated.
Over to dld for plan-B
- probability, discrete, continuous, cumulative, joint, independent, dependent, marginal, Bayes, (max-) likelihood, prior, posterior, Kullback-Leibler distance
- inference, model class ... model ... parameter(s), estimation, bias, invariance
- information, entropy
- (prefix-) codes, examples, Huffman, arithmetic coding
- binomial distribution
- multinomial distribution
- integer (codes for), geometric, log*, tree-code, other...
- normal distribution
- sequential data, low-order Markov models, Lempel-Ziv, PFSA etc.
- Fisher information matrix, general form of MML
- applications: segmentation, ...
- (piece-wise) straight line fitting, linear regression, polygon fitting
- sequences and approximate matching
- unsupervised classification, Snob
- supervised classification (decision-) trees and graphs
- ?image compression, Peter T?
- ?causal models, Kevin K?
D. Dowe & L. Allison © 1999,
Comp Sci and SWE, Monash University