Next: About this document
ICMS Workshop on the Vapnik-Chervonenkis Dimension
Programme (as at 5th September)
The following is a provisional programme for the Meeting, which takes
place from Monday 9th to Friday 13th September. The 10:00 orientation talk
on Monday will be at a superficial level -- providing no more than
signposts for the rest meeting -- and most participants will want
to join at the start of the meeting proper, at 11:00. All talks are
at 14 India Street.
Disclaimer:
this provisional timtable is likely to become less accurate as the week
progresses. In particular, it may be necessary to re-pack the contributed
talks to accommodate late offers; this may involve adjusting the times
given here by up to 10 minutes or so.
Monday
- 09:00 onwards
- Registration.
- 10:00
- A bluffers' guide to the V-C Dimension.
- 10:30-11:00
- Break.
- 11:00
- [Start of workshop proper.]
David Haussler (University of California at Santa Cruz),
VC Dimension, Covering Numbers and Worst-case Prediction
of Individual Sequences
- 12:10-15:00
- Lunch break.
- 15:00
-
Nicolò Cesa-Bianchi (University of Milan),
A graph-theoretic generalization of the Sauer-Shelah Lemma.
- 15:40
- David Steinsaltz (Technische Universität, Berlin),
Vapnik-Cervonenkis techniques for
estimating the maxima of certain stochastic processes.
- 16:20
- Shai Ben-David (Technion),
Ideals over the reals defined by notions of shattering.
- 17:00
- End.
Tuesday
- 09:50
- Anselm Blumer (Tufts University, MA),
Tutorial on PAC Learning.
- 11:00-11:30
- Break.
- 11:30
- Martin Anthony (London School of Economics),
Probabilistic generalisation of functions, and uniform convergence.
- 12:10-15:40
- Lunch break.
- 15:40
- John Shawe-Taylor (Royal Holloway College, London),
The role of the fat shattering dimension in concept learning.
- 16:20
- Michael Schmitt (Technische Universität, Graz),
On the VC-dimension of networks of spiking neurons.
- 17:00
- David Ritchie (James Clerk Maxwell Foundation),
James Clerk Maxwell.
- 18:00
- Reception.
[Please turn over.]
Wednesday
- 09:50
- Mark Jerrum (University of Edinburgh),
General upper bounds for VC dimension
in terms of descriptive and computational complexity.
- 10:20
- Leonid Gurvits (NEC Research Institute, Princeton),
Combinatorial and topological properties
of VC-dimension and its generalizations.
- 11:00-11:30
- Break.
- 11:30
- Angus Macintyre (Merton College, Oxford),
The connection between model theory and VC dimension.
- 12:10
- End.
Thursday
- 09:50
- Shai Ben-David (Technion),
VC-dimension, sample-compression schemes and learning.
- 11:00-11:30
- Break.
- 11:30
-
Sean B. Holden (University College, London),
VC dimension and the sample complexity of error estimation techniques.
- 12:10-15:00
- Lunch break.
- 15:00
- Eli Shamir (Hebrew University, Jerusalem),
Dimension of query spaces and noise-robust learning.
- 15:40
- Paul Fischer (Universität Dortmund),
A power law for the sample size when learning in the presence of noise.
- 16:20
- Hans Ulrich Simon (Universität Dortmund),
Learning in the presence of malicious noise
beyond the information-theoretic barrier
- 17:00
- End.
Friday
- 9:50
- Jirí Matousek (Charles University, Prague),
Geometric applications: discrepancy and derandomization.
- 11:00-11:30
- Break.
- 11:30
- Peter L. Bartlett (Australian National University),
Pattern classification and scale-sensitive versions of the
VC-dimension for neural networks.
- 12:10
- Pascal Koiran (ENS, Lyon),
VC dimension of feedforward and recurrent neural networks.
- 12:50
- End.
Next: About this document
Mark Jerrum
Thu Sep 5 11:21:06 BST 1996