AI Paper Atlas

Research map / Learning theory and algorithms / Prediction and decision losses

A Theoretical Framework for Modular Learning of Robust Generative Models

Corinna Cortes, Mehryar Mohri, Yutao Zhong

ICML 2026 · Learning theory and algorithms › Prediction and decision losses · approach: Active learning and subset selection · 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
use a frozen-expert gate instead of retraining one monolithic model on the union of domains~20M-param modular model · Wiki/Code/FineWeb NLL (diff seed)NLL per token: 5.133 -> 4.994Improved
use a frozen-expert gate instead of retraining on the union (different data splits)~20M-param modular model · Wiki/Code/FineWeb NLL (diff data)NLL per token: 5.306 -> 5.087Improved
use a frozen-expert gate instead of a jointly trained token-level MoE with load balancing~20M-param modular model · Wiki/Code/FineWeb NLL (diff seed)NLL per token: 5.080 -> 4.994Improved
scale to 1.6B-parameter experts with a 2.5M-param gate versus a 4.9B monolithic modelRobust Gate · real-world data, 1B+ scaletest NLL: 4.829 -> 4.734Improved
use gate rather than fixed-mixture retrained models, including larger onesRobust Gate · synthetic contradictory domains (+1/-1 rules)NLL per tokenImproved
compare gate with oracle models retrained on exact test mixtureRobust Gate · synthetic taskNLL per tokenMixed
distill non-causal gate into a causal router (Structural Distillation)Causal Router (d=10) · synthetic taskNLL per tokenImproved
increase monolithic baseline capacity to 1.5x (ablation)Retrained Fixed model · synthetic contradictory domainsNLL per tokenNo clear effect
use Lagrangian primal-dual instead of quadratic penalty for normalization (ablation)Gate training · algorithm designconstraint satisfactionImproved

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