Главная
Study mode:
on
1
Intro
2
(Motivation) Dynamic Pricing
3
Bandits w/ Knapsacks (BWK)
4
Prior Work - Stochastic BwK
5
Background: Feedback Models
6
Main Result
7
Why is BwK hard?
8
Why is Adversarial BwK harder?
9
Benchmark
10
Overview
11
Linear Relaxation
12
Lagrange Game
13
a: Main algorithm (MAIN)
14
Step 3b: Learning in Games
15
Regret Bound
16
Challenges
17
Simple Algorithm
18
High-prob. v/s Adaptive Adversary
19
Extensions
20
Future Work
Description:
Explore the concept of Adversarial Bandits with Knapsacks in this 21-minute IEEE conference talk. Delve into dynamic pricing, prior work on Stochastic BwK, and various feedback models. Examine the main results, challenges, and reasons behind the complexity of BwK and Adversarial BwK. Learn about linear relaxation, Lagrange games, and the main algorithm (MAIN). Discover regret bounds, simple algorithms, and the differences between high-probability and adaptive adversary scenarios. Conclude with extensions and potential future work in this field.

Adversarial Bandits with Knapsacks

IEEE
Add to list