AI Paper Atlas

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

Latent Spherical Flow Policy for Reinforcement Learning with Combinatorial Actions

Lingkai Kong, Anagha Satish, Hezi Jiang, Akseli Kangaslahti, Andrew Ma, Wenbo Chen, Mingxiao Song, Lily Xu, Milind Tambe

ICML 2026 · Reinforcement learning › Policy gradient and actor-critic · approach: Model-based RL with guarantees

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 stochastic latent spherical flow policy instead of deterministic structured policy (SRL)LSFlow · 4 combinatorial RL tasksaverage return: 22.91 -> 27.62Improved
train flow policy in latent cost space with one solver call per step instead of differentiating through solver (SRL)LSFlow · 4 combinatorial RL tasksaverage training time (h): 3.89 -> 1.29Improved
add spherical flow matching in place of Euclidean flow (ablation)LSFlow · Dynamic Scheduling and Dynamic RoutingreturnImproved
add moderate vMF smoothing of the Bellman operator vs none or too strong (ablation)LSFlow · combinatorial RL tasksreturnImproved
train critic in latent cost space instead of action space (ablation)LSFlow · Dynamic Schedulingtime per policy update: 6 minutes -> under 1 secondImproved
add LSFlow instead of SEQUOIA's MIP-embedded Q-networkLSFlow · Dynamic Schedulingreturn: 24.00 -> 28.85Improved
use LSFlow vs SRL on STI testing networksLSFlow · STI testing (4 disease networks)fraction of positives detectedMixed

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