Fixed-point Error Bounds for Mean-payoff Markov Decision Processes
A Google TechTalks, presented by Roberto Cominneti, 2024-03-19
A Google Algorithms Seminar. ABSTRACT: We discuss the use of optimal transport techniques to derive finite-time error bounds for reinforcement learning in mean-payoff Markov decision processes. The results are obtained as a special case of stochastic Krasnoselski—Mann fixed point iterations for nonexpansive maps. We present sufficient conditions on the stochastic noise and stepsizes that guarantee almost sure convergence of the iterates towards a fixed point, as well as non-asymptotic error bounds and convergence rates. Our main results concern the case of a martingale difference noise with variances that can possibly grow unbounded. We also analyze the case of uniformly bounded variances, and how they apply for Stochastic Gradient Descent in convex optimization.
ABOUT THE SPEAKER: Roberto Cominetti is a professor with the Faculty of Engineering and Sciences at Universidad Adolfo Ibáñez, Santiago, Chile. His research interests include convex analysis and game theory and their applications in transportation networks.
1 view
0
0
2 months ago 00:01:35 1
Silent Hill 3 - Modern Camera Mod Release (Third Person + First Person)
2 months ago 00:02:35 4
[Fix] Bsod When Updating BIOS on Windows 10
3 months ago 00:29:03 1
We Dueled With the NEW Meta Decks After the Ban List! | Duel Challenge!
5 months ago 00:06:06 1
Learn CORS In 6 Minutes
5 months ago 00:04:00 1
How to Fix missing error(DISCORD) | Entry Point Error| FIX Windows 7| 2020
5 months ago 00:05:41 1
Pixel Verse Airdrop: How to create pixel Tap dashboard and connect wallet using bybit web3 #airdrop
6 months ago 00:21:56 1
A First Look At SparkyLinux 7.4
7 months ago 00:07:29 1
I also make MISTAKES!
8 months ago 00:08:54 1
Helldivers 2 - How to FIX FPS Drops and 0 Input Latency on ANY PC!✅
9 months ago 00:57:54 1
Fixed-point Error Bounds for Mean-payoff Markov Decision Processes
10 months ago 00:08:41 1
Turning a broken 2 ton robot into a CNC-machine | ABB IRB6400
10 months ago 00:04:08 1
InfinityBox CM2MT2 Latest Boot 2023 Secuirty Work all Old Version Without Reactivation or Update
10 months ago 00:14:17 1
Oppo A83T(A1) China touch not work Error charger/Firmware troubled after flashing FIX Done 100%
10 months ago 00:06:58 1
VAN 9003 Valorant Windows 11 Error Fixed | This Build of Vanguard is Out of Compliance [2023]
11 months ago 00:11:53 1
4 PowerPoint MISTAKES To AVOID in 2024!
11 months ago 00:20:19 2
Tantei Opera Milky Holmes Eng Patched with Old W7OS using PPSSPP v1.3 ZD SC 7 Full RC I Part 2
1 year ago 00:04:01 1
【Ken-chan, URAY, Alex M., VinnyO, & Dav-P】Just Monika (Random Encounters - REDONE)【Genderbent Cover】
1 year ago 00:02:09 1
Nokia Tune History (1994-2015) - PLEASE WATCH UPDATED VIDEO, LINK IN DESCRIPTION
1 year ago 00:38:47 1
We Exploded the AMD Ryzen 7 7800X3D & Melted the Motherboard
1 year ago 00:35:53 1
“Marx & Satan“ by Richard Wurmbrand - Chapter 5
1 year ago 00:20:35 1
“Marx & Satan“ by Richard Wurmbrand - Intro & Chapter 1
1 year ago 00:01:11 5
Mushishi OST 1 - Kotowari
1 year ago 00:36:47 1
Blender Photogrammetry: Repairing Holes & Fix Bad Meshes in Photo-Scan Models