AI Paper Atlas

Research map / Large language models / Agents, code and math

Simulation-Based Approach to Efficient Commonsense Reasoning in Very Large Knowledge Bases

Abhishek Sharma, Keith M. Goolsbey

AAAI 2019 · Large language models › Agents, code and math · approach: Inductive and abstract reasoning evaluation · 2 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 Cyc default heuristic policy with MCTS-learned policyCyc inference engine · Experiment 1% queries answered: 46 -> 92Improved
replace Cyc default heuristic policy with MCTS-learned policyCyc inference engine · Experiment 1speedup: 7.1Improved
replace default policy with MCTS policyCyc inference engine · Experiment 2% queries answered: 66 -> 94Improved
replace default policy with MCTS policyCyc inference engine · Experiment 3% queries answered: 76 -> 94Improved
replace default policy with MCTS policyCyc inference engine · Experiment 4% queries answered: 82 -> 94Improved
replace default policy with MCTS policyCyc inference engine · Experiment 4speedup: 3.3Improved
use MCTS policy on easy queries (0-100 s in baseline)Cyc inference engine · 597 easiest queriesspeedup: 1.23No clear effect
larger exploration constant c = 1000 at ~100k simulations (ablation)MCTS policy · Cyc queriesspeedup / Q/AImproved
increase number of simulations (ablation)MCTS policy · Cyc queriesspeedup / Q/ANo clear effect
use small to medium c (10 or 100) with more simulations (ablation)MCTS policy · Cyc queriesspeedup / Q/AImproved

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