Veranstaltungen
Diese Woche
×
Modal title
Modal content
Montag, 3. Juli | |||
---|---|---|---|
Zeit | Referent:in | Titel | Ort |
15:15 - 16:15 |
Prof. Dr. Jim Daicall_made Cornell University and The Chinese University of Hong Kong, Shenzhen |
Abstract
Conservative update methods such as Trust Region policy optimization and Proximal policy optimization (PPO) have become the dominant reinforcement learning algorithms because of their ease of implementation and good practical performance. A conventional setup for notoriously difficult queueing network control problems is a Markov decision problem (MDP) that has three features: infinite state space, unbounded costs, and long-run average cost objective. We extend the theoretical framework of these conservative update methods for such MDP problems. The resulting PPO algorithm is tested on a parallel-server system and large-size multiclass queueing networks. The algorithm consistently generates control policies that outperform state-of-art heuristics in literature in a variety of load conditions from light to heavy traffic. These policies are demonstrated to be near-optimal when the optimal policy can be computed. A key to the successes of our PPO algorithm is the use of three variance reduction techniques in estimating the relative value function via sampling. First, we use a discounted relative value function as an approximation of the relative value function. Second, we propose regenerative simulation to estimate the discounted relative value function. Finally, we incorporate the approximating martingale-process method into the regenerative estimator. This is joint work with Mark Gluzman at Meta.
Talks in Financial and Insurance MathematicsMore information: https://math.ethz.ch/imsf/courses/talks-in-imsf.htmlcall_made Queueing Network Controls via Deep Reinforcement Learningread_more |
HG G 43 |
15:15 - 16:15 |
Prof. Dr. Jim Daicall_made Cornell University and The Chinese University of Hong Kong, Shenzhen |
Abstract
For more information on this seminar see page ``Talks in Financial and Insurance Mathematics'' at https://math.ethz.ch/imsf/courses/talks-in-imsf.html?s=fs23
DACO SeminarMore information: https://math.ethz.ch/imsf/courses/talks-in-imsf.html?s=fs23call_made Queueing Network Controls via Deep Reinforcement Learning (Financial and Insurance Mathematics seminar, cross-listed)read_more |
HG G 43 |
Dienstag, 4. Juli | |||
---|---|---|---|
— keine Veranstaltungen geplant — |
Mittwoch, 5. Juli | |||
---|---|---|---|
Zeit | Referent:in | Titel | Ort |
15:00 - 16:30 |
Georg Anegg Examiner: Prof. Dr. Rico Zenklusen |
Abstract
Techniques for Designing Approximation Algorithms for Fair Clustering Problems |
HG G 19.1 |
Donnerstag, 6. Juli | |||
---|---|---|---|
— keine Veranstaltungen geplant — |
Freitag, 7. Juli | |||
---|---|---|---|
— keine Veranstaltungen geplant — |