Preliminary Topic list
These are my notes from our initial meeting for topics me might want to cover. You can compare it to last year's.
NIPS highlights
Non-parametric Methods
- Pitmann lecture notes on stochastic processes.
- Heirarchical models: infinite HMM, heirarchical LDA.
Randomized Algorithms
- Motwani, Raghavan book
- MCMC from Computer Science background. Jerrum and Sinclair paper.
Advanced MCMC Sampling
- Regeneration
- Parallel Tempering
- Exact sampling on continuous or otherwise interesting systems.
Spam Filtering
- Non-naive Bayesian methods.
Linear Response
Wainwright and Jordan Review Paper
Frequentist methods(!)
- Hypothesis testing, Confidence intervals, p-values.
- Rank based methods.
Information Theory
Learning Theory
- Large Deviation Bounds.
- Online learning. Mistake bounds, Littlestone and Warmuth
Information Bottle-neck
Graphical Models
- Learning graph structure.
Game Theory
Manifold learning
- Multidimensional scaling
- Automatic discovery of intrinsic dimensionality
An Extracellular Darwinian experiment
Causality
- Pearl; Dawid; Dempster; Glymour et al.
Reinforcement Learning
- Bayesian decision theory.
- Multi-agent RL.