AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

A Deep Reinforcement Learning Framework for Column Generation

Cheng Chi, Amine Mohamed Aboussalah, Elias Boutros Khalil, Juyoung Wang, Zoha Sherkat-Masoumi

NeurIPS 2022 · Reinforcement learning › Visual and representation RL · approach: RL with tree search for discovery · 13 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 greedy column selection with RLCGcolumn generation · CSP n=750mean CG iterations: 292.8 -> 227.2Improved
replace greedy column selection with RLCGcolumn generation · CSP n=50mean CG iterations: 51.2 -> 41.6Improved
replace greedy column selection with RLCGcolumn generation · VRPTW largemean CG iterations: 128.1 -> 75.6Improved
replace greedy column selection with RLCGcolumn generation · VRPTW smallmean CG iterations: 18.8 -> 9.5Improved
replace MILP one-step-lookahead expert with RLCGcolumn generation · CSP n=750mean time (s): 827.9 -> 460.3Improved
replace MILP expert with RLCGcolumn generation · CSP n=50mean CG iterations: 42.5 -> 41.6Mixed
train with curriculum (easy to hard) rather than random order (ablation)RLCG · CSP trainingCG iterations during trainingImproved
weight objective decrease in reward (alpha=300 vs alpha=0) (ablation)RLCG · CSP validationgreedy/RL iteration ratio: 0.91 -> 1.25Improved
different random seeds for best model (ablation)RLCG · CSP validationgreedy/RL iteration ratio: 1.23 to 1.25No clear effect

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