AI Paper Atlas

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

Temporal Difference Calibration in Sequential Tasks: Application to Vision-Language-Action Models

Shelly Francis-Meretzki, Mirco Mutti, Yaniv Romano, Aviv Tamar

ICML 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Temporal-difference and discounting theory · 0 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
replace binary cross-entropy with TD loss (TDQC) (ablation)black-box RNN predictor · OpenVLA WidowX, unseen taskssequential Brier score: .344 -> .192Improved
replace binary cross-entropy with TD loss (TDQC) (ablation)black-box RNN predictor · UniVLA LIBERO, unseen tasksROC-AUC: 54.37 -> 70.17Improved
replace MC/BCE loss with TD loss (ablation)SAFE-RNN · pi0 LIBERO, unseen taskssequential Brier score: .172 -> .097Improved
replace MC/BCE loss with TD loss (ablation)SAFE-RNN · pi0-FAST Franka, unseen tasksROC-AUC: 56.69 -> 68.43Improved
replace MC/BCE loss with TD loss (ablation)SAFE-RNN · OpenVLA LIBERO, unseen tasksROC-AUC: 69.04 -> 65.12Hurt
replace BCE with TD loss (ablation)SAFE-MLP · OpenVLA LIBERO, unseen tasksROC-AUC: 64.99 -> 60.09Hurt
replace BCE with TD loss (ablation)SAFE-MLP · pi0-FAST LIBERO, unseen taskssequential Brier score: .162 -> .150Improved
replace BCE with TD loss (ablation)SAFE-MLP · OpenVLA LIBERO, unseen taskssequential Brier score: .231 -> .229No clear effect
train black-box predictor on action probabilities with TDQC instead of static heuristicsblack-box RNN · OpenVLA LIBERO, unseen taskssequential Brier score: .313 -> .197Improved
use TD-calibrated action probabilities (black box) instead of white-box hidden statesRNN-TDQC vs SAFE-RNN · OpenVLA LIBERO, unseen taskssequential Brier score: .255 -> .197Improved
use TDQC-trained value predictor to guide action search among 10 sampled actionsOpenVLA · LIBERO-10, 3 held-out taskssuccess rate: 13% over the regular baseline (55% average)Improved
use BCE-trained value predictor for guided action searchOpenVLA · LIBERO-10, 3 held-out taskssuccess rate: 50%Improved

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