Tag

ResAct

0 views collected around this technical thread.

Kuaishou Tech
Kuaishou Tech
Mar 29, 2023 · Artificial Intelligence

ResAct: A Reinforcement Learning Approach for Long-Term User Retention in Sequential Recommendation

The paper introduces ResAct, a reinforcement‑learning framework that improves long‑term user retention in sequential recommendation by constraining the policy space near the online‑serving policy and employing a conditional variational auto‑encoder, residual actor, and state‑action value network, achieving significant gains over existing methods on a large‑scale short‑video dataset.

Recommendation systemsResActreinforcement learning
0 likes · 9 min read
ResAct: A Reinforcement Learning Approach for Long-Term User Retention in Sequential Recommendation