Resource Allocation in Multi-armed Bandits by Kirthevasan Kandasamy

Resource Allocation in Multi-armed Bandits by Kirthevasan Kandasamy

Channel:
United States Google TechTalks
Subscribers:
348,000
Published on ● Video Link: https://www.youtube.com/watch?v=_hL4gMJrqXc


Duration: 59:48
1,402 views
0

A Google TechTalk, presented by Kirthevasan Kandasamy, 2021/06/09
ABSTRACT: Most bandit problems are formulated in terms of the number of arm pulls (sample complexity), where we usually wish to find the optimum of an unknown system while minimising the number of arm pulls. However, executing arm pulls requires resources, and in practice, we are faced with a variety of resource constraints when pulling arms.
In this talk, I will discuss some recent work that studies two settings in this topic. In the first, we are given a finite set of resources. We can choose to execute a variable number of arm pulls with these resources at a given time, but due to sublinear scaling of real world systems, we need to handle trade-offs between information accumulation and resource efficiency. In the second, we have an elastic resource (e.g. cloud computing) where there is no constraint on the amount of resources that can be used at a given time. However, we wish to minimise the total resource-time used while adhering to a strict deadline and performance requirement. I will discuss algorithms, theoretical upper/lower bounds, and empirical results on hyperparameter tuning.

About the speaker: Kirthevasan (Kirthi) Kandasamy is a postdoctoral scholar at UC Berkeley. He works on a variety of topics in the intersection of sequential decision-making under uncertainty, game theory, and systems.



Other Videos By Google TechTalks


2021-09-30A Regret Analysis of Bilateral Trade
2021-09-29CoinPress: Practical Private Mean and Covariance Estimation
2021-09-29"I need a better description": An Investigation Into User Expectations For Differential Privacy
2021-09-29On the Convergence of Deep Learning with Differential Privacy
2021-09-29A Geometric View on Private Gradient-Based Optimization
2021-09-29BB84: Quantum Protected Cryptography
2021-09-29Fast and Memory Efficient Differentially Private-SGD via JL Projections
2021-09-29Leveraging Public Data for Practical Synthetic Data Generation
2021-07-13Efficient Exploration in Bayesian Optimization – Optimism and Beyond by Andreas Krause
2021-07-13Learning to Explore in Molecule Space by Yoshua Bengio
2021-07-13Resource Allocation in Multi-armed Bandits by Kirthevasan Kandasamy
2021-07-13Grey-box Bayesian Optimization by Peter Frazier
2021-06-10Is There a Mathematical Model of the Mind? (Panel Discussion)
2021-06-04Dataset Poisoning on the Industrial Scale
2021-06-04Towards Training Provably Private Models via Federated Learning in Practice
2021-06-04Breaking the Communication-Privacy-Accuracy Trilemma
2021-06-04Cronus: Robust Knowledge Transfer for Federated Learning
2021-06-04Private Algorithms with Minimal Space
2021-06-04Tackling the Objective Inconsistency Problem in Heterogeneous Federated Optimization
2021-06-04Flower: A Friendly Federated Learning Framework
2021-06-04Lions, Skunks, and Kangaroos: Geo-Distributed Learning on the Flickr-Mammal Dataset