CSE454/Local/2003/

When and where: weeks 1-13, semester 1, 3-5pm Tuesdays, E3. There will be a 1-hour lecture 3-4, and occasional Q+A or "tute" after.

NB. You will need an account on the csse machine `nexus' to do prac2.

Subject to change...

• week 1, Tuesday 4 March
  [Admin] [Intro]
  Reading: [kissing] [fair Euro?] [dodgy dice]
• week 2, Tuesday 11 March
  ([Data & models and/or) [ACSC-2003] ...
• week 3, Tuesday 18 March
  ... ACSC-2003, [information] [Snob]
  Reading: [ACSC-2003]
  also prac1 and Snob, see below!
• week 4, Tuesday 25 March
  [Unsupervised class'n] [finite-state] [2-state'n](methods 1&2)...
  Reading: [SMML] not examinable, but
  [2-state] & [m-state] are
• week 5, Tuesday 1 April
  ...2-state (method-3). Use 0th- and 1st-order Markov models [here], model complexity v. fit to data.
  [Normal] distribution.
  Reading: [Normal]
• week 6, Tuesday 8 April
  [Mixture] modelling and [algorithm]
• week 7, Tuesday 15 April
  [Supervised class'n] [Classification Trees1] ...

  -- Easter --

• week 8, Tuesday 29 April
  ... [Classification Trees1] ...

  [prac1] 20% due noon Thurs, week 8, 1 May *[notes] on prac1, 12/5/2003

  
  
• week 9, Tuesday 6 May
  ... [Classification Trees2] talked about prac1, so only got to "tree search problem"
  [algorithm] and an [application] ...
  Please form a consensus
  on an exam data and time
  (pref' in week 14+)
• week 10, Tuesday 13 May
  ...(finish above) [Fisher info]-(carry-over)
• week 11, Tuesday 20 May
  ... [multi-state] ...
  ...also read [N(m,s) Fisher]
• week 12, Tuesday 27 May
  ... and the multi-state estimator.
  [Int codes], [Coding] including a sketch(!) of arithmetic coding.

  [prac2] 20% due noon Fri, week 12, 30 May

  
  
• week 13, Tuesday 3 June
  summary:

  building block (distribution)	examples of structured models
  	cond' prob' table	order_k Markov model	prob' finite state automaton PFSA/ HMM	mixture model	class'n tree
  an int' code		k (unless common knowledge)	# states	# components (classes)
  multi state	for each "row"	for each "context"	(i)for arc existence (ii)exit probabilities (per state)	(i)abundances of components (ii)discrete attributes	leaf categories
  continuous, e.g. N(m,s)				continuous attributes	(leaf distributions, if allowed)
  other?					structure of tree

  
  
• Examination: 60%, Date to be decided:
  Room G27, building 72 (seats 25), Friday 13 June, 2003, 2pm-4pm

Past examination papers: [2003] (after the event), [2002], and other [possible questions].