Seven papers are accepted to KSC 2023. Check our papers (click).


κ°•ν™”ν•™μŠ΅ / ν”„λ ˆμž„μ›Œν¬

  • ARCLE: 좔상화 및 좔둠을 μœ„ν•œ κ°•ν™”ν•™μŠ΅ ν™˜κ²½ (μš°μˆ˜λ°œν‘œλ…Όλ¬Έμƒ μˆ˜μƒ)
  • μ›”λ“œ λͺ¨λΈμ„ μ΄μš©ν•œ ARC 문제의 사전 지식 평가
  • μ›”λ“œλͺ¨λΈμ„ ν†΅ν•œ ARC의 핡심 지식 μΆ”μΆœ

기계 ν•™μŠ΅ / 방법둠

  • κ·€λ‚© 편ν–₯ μ œκ³΅μ„ μœ„ν•œ 색채 μ–΄ν…μ…˜ ν•™μŠ΅ (μš°μˆ˜λ…Όλ¬Έμƒ μˆ˜μƒ)
  • 좔상화 및 μΆ”λ‘  문제 해결을 μœ„ν•œ λŒ€μ‘°ν•™μŠ΅

데이터셋 / λŒ€ν˜• μ–Έμ–΄ λͺ¨λΈ

  • λŒ€ν˜• μ–Έμ–΄ λͺ¨λΈμ„ ν™œμš©ν•œ 퓨샷 μΆ”λ‘  문제의 데이터 증강 (μš°μˆ˜λ°œν‘œλ…Όλ¬Έμƒ μˆ˜μƒ)
  • κ±°λŒ€μ–Έμ–΄λͺ¨λΈμ˜ μΆ”λ‘ λŠ₯λ ₯ 평가λ₯Ό μœ„ν•œ MC-LARC 데이터셋

Reinforcement Learning / Framework

  • ARCLE: Gymnasium Environment for Abstraction and Reasoning Corpus
  • Evaluating Prior Knowledge of ARC Using World Models
  • Extracting the Core Knowledge of ARC with the World Model

Machine Learning / Methodology

  • Color Attention for Inductive Bias Provision
  • Using Contrastive Learning for Abstraction and Reasoning Task

Dataset / Large Language Model

  • Augmenting Few-Shot Demonstrations with Large Language Model
  • MC-LARC Benchmark to Measure LLM Reasoning Capability

Group

Jiwon