Sejin Kim
I am a Postdoctoral Researcher at the Department of AI Convergence, Gwangju Institute of Science and Technology (GIST), Korea. I received my Ph.D. in Computer Science from KAIST, where I worked under the supervision of Prof. Jae-Gil Lee, focusing on spatio-temporal prediction and transfer learning for data-poor cities.
My current research focuses on human-like AI, artificial general intelligence, reinforcement learning, and program synthesis. I have published in leading venues, including NeurIPS, KDD, ICML, and IJCAI. I am the recipient of the NRF Postdoctoral Fellowship (2024–2026) for developing AI models with advanced abstraction and reasoning capabilities. I was recognized with the Excellence Postdoctoral Researcher Award at GIST in 2024 and selected as an Outstanding Reviewer at KDD 2025.
Publications
ARCTraj: A Dataset and Benchmark of Human Reasoning Trajectories for Abstract Problem Solving
Kim, S., Choi, H., Lee, S., & Kim, S. (2026). ARCTraj: A Dataset and Benchmark of Human Reasoning Trajectories for Abstract Problem Solving. KDD Datasets and Benchmarks. (top conference)
Composed Program Induction with Latent Program Lattice
Park, J., Park, J., Shim, J., Kim, S., Vennemann, P., & Kim, S. (2025). Composed Program Induction with Latent Program Lattice. NeurIPS Workshop NeurReps.
Reasoning Abilities of Large Language Models: In-Depth Analysis on the Abstraction and Reasoning Corpus
Lee, S., Sim, W., Shin, D., Hwang, S., Seo, W., Park, J., Lee, S., Kim, S., & Kim, S. (2025). Reasoning Abilities of Large Language Models: In-Depth Analysis on the Abstraction and Reasoning Corpus. ACM TIST 16(6).
Diffusion-model-guided Implicit Q-learning with Adaptive Revaluation
Park, J., Kim, Y., Kim, S., Lee, B., & Kim, S. (2025). Diffusion-model-guided Implicit Q-learning with Adaptive Revaluation. CoRL Workshop LEAP.
Solution Augmentation for ARC-AGI Problems Using GFlowNet: A Probabilistic Exploration Approach
Hwang, S., Lee, S., Kim, S., & Kim, S. (2025). Solution Augmentation for ARC-AGI Problems Using GFlowNet: A Probabilistic Exploration Approach. TMLR.
Addressing and Visualizing Misalignments in Human Task-Solving Trajectories
Kim, S., Lee, H., & Kim, S. (2025). Addressing and Visualizing Misalignments in Human Task-Solving Trajectories. KDD. (top conference, acceptance rate: 18.4%)
TRACED: Transition-aware Regret Approximation with Co-learnability for Environment Design
Cho, G., Im, J., Lee, J., Yi, H., Kim, S., & Kim, S. (2025). TRACED: Transition-aware Regret Approximation with Co-learnability for Environment Design. arXiv.
Addressing and Visualizing Misalignments in Human Task-Solving Trajectories
Kim, S., Lee, H., & Kim, S. (2025). Addressing and Visualizing Misalignments in Human Task-Solving Trajectories. ICLR Workshop BiAlign.
System-2 Reasoning via Generality and Adaptation
Kim, S. & Kim, S. (2024). System-2 Reasoning via Generality and Adaptation. NeurIPS Workshop S2R.
Reasoning Abilities of Large Language Models through the Lens of Abstraction and Reasoning
Lee, S., Sim, W., Shin, D., Kim, S., & Kim, S. (2024). Reasoning Abilities of Large Language Models through the Lens of Abstraction and Reasoning. NeurIPS Workshop S2R.
Diffusion-Based Offline RL for Improved Decision-Making in Augmented ARC Task
Kim, Y., Park, J., Kim, H., Kim, S., Lee, B., & Kim, S. (2024). Diffusion-Based Offline RL for Improved Decision-Making in Augmented ARC Task. arXiv.
O2ARC 3.0: A Platform for Solving and Creating ARC Tasks
Shim, S., Ko, D., Lee, H., Lee, S., Song, D., Hwang, S., Kim, S., & Kim, S. (2024). O2ARC 3.0: A Platform for Solving and Creating ARC Tasks. IJCAI Demo.
Enhancing Analogical Reasoning in the Abstraction and Reasoning Corpus via Model-Based RL
Lee, J., Sim, W., Kim, S., & Kim, S. (2024). Enhancing Analogical Reasoning in the Abstraction and Reasoning Corpus via Model-Based RL. IJCAI Workshop IARML.
ARCLE: The Abstraction and Reasoning Corpus Learning Environment for Reinforcement Learning
Lee, H., Kim, S., Lee, S., Hwang, S., Lee, J., Lee, B., & Kim, S. (2024). ARCLE: The Abstraction and Reasoning Corpus Learning Environment for Reinforcement Learning. CoLLAs.
ARC 문제 해결을 위한 프롬프트 엔지니어링의 가능성
Sim, W., Jin, H., Kim, S., & Kim, S. (2023). The Possibility of Prompt Engineering for ARC Problem Solving. KIISE Transactions on Computing Practices.
월드모델을 통한 ARC의 핵심 지식 추출
Lee. J., Lee, S., Kim, S., & Kim, S. (2023). Extracting the Core Knowledge of ARC with the World Model. Korea Software Congress.
추상화 및 추론 문제 해결을 위한 대조학습
Gu, G., Sim, W., Im, J., Kim, S., & Kim, S. (2023). Using Contrastive Learning for Abstraction and Reasoning Task. Korea Software Congress.
ARCLE: 추상화 및 추론을 위한 강화학습 환경
Lee. H., Kim, S., & Kim, S. (2023). ARCLE: Gymnasium Environment for Abstraction and Reasoning Corpus. Korea Software Congress.
Unraveling the ARC Puzzle: Mimicking Human Solutions with Object-Centric Decision Transformer
Park, J., Im, J., Hwang, S., Lim, M., Ualibekova, S., Kim, S., & Kim, S. (2023). Unraveling the ARC Puzzle: Mimicking Human Solutions with Object-Centric Decision Transformer. ICML Workshop ILHF.
의사결정 트랜스포머를 사용한 추상화와 추론
Park., J., Im., J., Lee, Y., Shin, D., Kim, S., & Kim, S. (2023). Abstraction and Reasoning Challenge with Decision Transformer. Korea Computer Congress.
준지도 시계열 분류를 위한 세그먼트 대조학습
Shin, Y., Park, D., Kang, J., & Kim, S. (2022). Contrastive Learning with Segment Supervision for Semi-Supervised Time Series Classification. Korea Software Congress.
Autoencoder ODEs for Time Series Anomaly Detection
Kim, B., Na, J., Byun, Y., Kim, S., & Lee., J.-G. (2022). Autoencoder ODEs for Time Series Anomaly Detection. Korea Software Congress.
AWS DeepRacer 를 활용한 라이다 센서 유무에 따른 강화학습 기반 자율주행 모델 성능 비교
Kim, J., Kim, D., Kim, S., Kang, J., Park, D., & Lee., J.-G. (2022). Performance Comparison of Self-Driving Models With and Without LIDAR Sensors Based on Reinforcement Learning in AWS DeepRacer Platform . Korea Computer Congress.
Generating a Machine Learning Model with a Few Sentences
Trirat, P., Shin, Y., Kim, S., & Kim, M. (2021). Generating a Machine Learning Model with a Few Sentences. Korea Software Congress.
Revisit Prediction by Deep Survival Analysis
Kim, S., Song, H., Kim, S., & Lee, J.-G. (2020). Revisit Prediction by Deep Survival Analysis. PAKDD.
Building Social Networking Services Systems Using the Relational Shared-Nothing Parallel DBMS
Whang, K.Y., Na, I., Yun, T., Park, J., Cho, K., Kim, S., Yi, I., & Lee, B. (2020). Building Social Networking Services Systems Using the Relational Shared-Nothing Parallel DBMS. DKE.
거시적 동선 정보에 기반한 고객 재방문 예측
Kim, B., Kim, S., Kim, S., & Lee, J.-G. (2019). Customer Revisit Prediction Using Macroscale Mobility Information. Korea Software Congress.
Spatial OLAP 에서 공간 연산자의 구현 기법 비교
Kim, S. (2018). A Comparison of Implementation Method of Spatial Operator in Spatial OLAP. Korea Computer Congress.
SentiWorld: Understanding Emotions between Countries Based on Tweets
Yea, S.-J., Kim, S., TO, J.-M., & Lee, J.-G. (2016). SentiWorld: Understanding Emotions between Countries Based on Tweets. ICWSM Demo.