Bayesian Bandits (Part 2)
eigenfoo opened this issue · 0 comments
eigenfoo commented
- Walk through two extensions to the traditional Thompson sampling algorithm presented in my first blog post.
- Non-stationary bandits
- Decaying posteriors
- Application: use Thompson sampling & decaying posteriors for portfolio construction / algorithm selection?
- Contextual bandits
- Non-stationary bandits
- Miscellaneous
Some resources: