AI Paper Atlas

Research map / Reinforcement learning / Policy gradient and actor-critic

Robust Losses for Learning Value Functions

Andrew P. Patterson, Victor Liao, Martha White

TPAMI 2022 · Reinforcement learning › Policy gradient and actor-critic · approach: Temporal-difference and discounting theory · 21 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use Huber Bellman error instead of squared Bellman error (MHBE vs MSBE fixed point)linear value function · HardAlias-2value errorImproved
use robust objectives instead of MSBE on idealized chainslinear value function · SmallChain/BigChainvalue errorNo clear effect
use L1 gradient-correction (L1-TDC) instead of GTD2 variantgradient TD · linear predictionMSVEHurt
use robust TDC variants instead of squared TDCgradient TD · Baird counterexampleMSVEImproved
use QRC-Huber instead of QRC (squared loss)QRC · CliffWorldaverage returnImproved
use QRC-Huber instead of DQNDQN · classic control (5 domains)average returnImproved
add target networks to QRC and QRC-Huber (ablation)QRC-Huber · classic controlaverage returnHurt
remove target networks from DQN (ablation)DQN · Cart-poleaverage returnHurt
use QRC-Huber instead of QRCQRC · Minatarscaled returnImproved
use QRC-Huber instead of DQNDQN · Minatarscaled returnImproved
use saddlepoint GQ instead of gradient-correction QRC-Huber (ablation)nonlinear control · classic controlmean returnHurt
use smaller Huber threshold tau (ablation)QRC-Huber · Mountain Carperformance distributionHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy