News & Updates

Invited Tutorial Speaker at Astro AI Workshop 2025: 'Exploring Compositional Generalization of Neural Networks through Synthetic Experiments'.

Invited Speaker at Prague Synapse 2025: 'New News: System-2 Fine-tuning for Robust Integration of New Knowledge'. Slides.

Talk at NVIDIA: 'Decomposing Elements of Problem Solving: What "Math" does RL Teach?' Slides.

Joined Sigma-Xi, the scientific research honor society.

Seminar talk at KAIST Department of Physics: 'In-Context Learning: From toy models to practice.' Slides.

Seminar talk at University of Tokyo Department of Physics: 'In-Context Learning: Algorithms and Representations'. Slides.

Ph.D. Defense: 'Deep Learning as a Scientific Tool and a Model Organism of Intelligence'. Slides, Thesis.

Seminar talk at Marks and Sander Lab, Harvard Medical School: 'Fundamental Abilities and In-abilities of AI'. Slides.

I gave a talk on 'Understanding Fundamental Abilities of AI with synthetic experiments' at the Astro AI Lunch Seminar at Harvard Smithsonian Center for Astrophysics. Video.

Research internship at NTT Research on Understanding Mechanisms and Capabilities of AI with Hidenori Tanaka.

I will be presenting some of our works at NeurIPS 2024: (1) Our work on concept learning and hidden emergence of compositional capabilities will be presented as a spotlight poster! (2) Our work on algorithmic phases of in-context learning will be presented at the Workshop on Scientific Methods for Understanding Deep Learning. (3) Our work on how LLMs can reorganize their representations in a context dependent way will be presented at the Workshop on Symmetry and Geometry in Neural Representations.

Our paper 'Competition Dynamics Shape Algorithmic Phases of In-Context Learning' is on ArXiv!

CBS-NTT Fellow Candidate Talk: 'Towards a Neuroethology of AI: AI as a model system of intelligent phenomena'. Video, Slides.

Talk at Stanford Institute for Theoretical Physics: 'Understanding Compositional Generalization with Synthetic Data'. Slides.

Talk at Insight+Interaction Lab at Harvard: 'Understanding Compositional Generalization with Synthetic Data'. Slides.

Our paper 'Dynamics of Concept Learning and Compositional Generalization' is on ArXiv!

Our paper 'Emergence of Hidden Capabilities: Exploring Learning Dynamics in Concept Space' is accepted to NeurIPS 2024 as a Spotlight!

Poster at New England Mechanistic Interpretability Workshop 2024: 'Emergence of In-Context Learning Beyond Bayesian retrieval: A mechanistic study'. Poster.

NTT Physics & Informatics Laboratory Journal Club: 'Scaling and In-Context Learning of Large Language Models'. Slides.

Tunadorable made a video about our paper: 'Emergence of Hidden Capabilities: Exploring Learning Dynamics in Concept Space'

I gave a talk about our project diff4stats at EAS 2024: 'Probabilistic Completion of Astrophysical Fields for Robust Statistics with Diffusion Models' slides

Summer research internship at NTT Research working on Understanding Mechanisms and Capabilities of AI.

Our paper 'Emergence of Hidden Capabilities: Exploring Learning Dynamics in Concept Space' from our project concept_learning is on arxiv.

I gave a talk about our project vdm4cdm_3d at the AstroAI Workshop 2024: '3D probabilistic reconstruction of the local dark matter from galaxies'

A paper from our project vdm4cdm_3d has been accepted at ICML 2024 Workshop Ai4Science: '3D Reconstruction of Dark Matter Fields with Diffusion Models: Towards Application to Galaxy Surveys'

A paper from our project concept_learning has been accepted at ICML 2024 Workshop on High-dimensional Learning Dynamics: 'Hidden Learning Dynamics of Capability before Behavior in Diffusion Models'

Our follow up paper of our project vdm4cdm_2d is on arxiv.

I gave a talk about our project vdm4cdm_2d at the ITC Luncheon: 'Debiasing with Diffusion: Probabilistic reconstruction of Dark Matter field from galaxies'

I gave a quick presentation of our project vdm4cdm_3d on using Diffusion Models to generate dark matter fields at the workshop 'AI-driven Discovery in Physics and Astrophysics' at Kavli IPMU.

I will be presenting two works at NeurIPS 2023 Workshops.

Our work tracking individual neurons in C.elegans using synthetically augmented data is published in Nature Methods.

Our project SmartEM, on optimizing neural imaging with real time deep learning is on the MIT News.

Won 2nd place at the Citadel Datathon competition.

I gave a talk about our project vdm4cdm_2d at the AstroAI lunch talk.

The paper about our project SmartEM is on bioarxiv.

Camels Virtual Telecoms hosted by the Flatiron Institute: 'Mstar2Mcdm using Diffusion Models: Generating Cold Dark Matter density fields'. Slides.

I will be presenting our project SmartEM at the ICML 2023 Workshop on Computational Biology.

I will present our project Whole Brain Imaging at the International C.elegans conference 2023.

Our work on constructing a galactic dustmap with minimal impact on cosmology is published in the Astrophysical Journal.

Our work on quantifying the effect of high-dimensional non-Gaussianity to statistical probes in cosmology is published in the Astrophysical Journal.

I will present our project targettrack at COSYNE 2023.

I will present our project targettrack at APS March Meeting 2023.

American Physical Society March Meeting 2023: 'Automated neuron tracking using deep learning and targeted augmentation'.

I will be presenting our work at AAS240.

American Astronomical Society 240: 'On the Gaussianity of Non-Gaussian probes of Large Scale Structure'.

Teaching Fellow for AP50: Physics as a Foundation for Science and Engineering with Dr. Eric Mazur at Harvard University.

Our work on quantitatively investigating latent spaces of Variational Autoencoders is on arxiv.

Our work on 3D imaging a nervous system in a naturalistic context is published in Cell00998-3). See the project Whole Brain Imaging.

Teaching Fellow for Physics 141: The Physics of Sensory Systems in Biology with Dr. Aravinthan Samuel at Harvard University.

Started as Research Assistant at Harvard University with Dr. Aravinthan Samuel, Dr. Douglas Finkbeiner, and Dr. Cecilia Garraffo.

I'm excited to start my Ph.D. program at Harvard Physics.

Received the Purcell Fellowship from Harvard University for 2019-2020.

Awarded the Doctoral Study Abroad Scholarship from KFAS (Korea Foundation for Advanced Studies).

Graduated Summa Cum Laude from KAIST with B.S. in Physics (Advanced Major), GPA: 4.08/4.3.

Received Best Machine Learning Project Award from KIAS.

Teaching Assistant for Physics Lab I at Seoul National University.

Received Physics Department Honorary Scholarship from KAIST.

Received Best Project Award at Physics Winter Camp from KIAS.

Stockholm International Youth Science Seminar 2018: 'Dark Matter Axion Search Experiment using 18T HTS Magnet'. Video.

Korean Physical Society Meeting 2018 Poster: 'Data acquisition system for dark matter Axion search experiment using 18 T HTS magnet at CAPP/IBS'.

General Physics Tutor for General Physics 2 at KAIST.

General Physics Tutor for General Physics 1 at KAIST.

Dean's List, KAIST Physics Department.

Exchange student at École Polytechnique in Palaiseau, France.

Awarded Undergraduate Student Scholarship from KFAS (2017-2019).

General Physics Tutor for General Physics 1 at KAIST.

Received Best Buddy Award from KAIST International Office.

General Physics Tutor for General Physics 2 at KAIST.

General Physics Tutor for General Physics 1 at KAIST.

Fall Dean's List at KAIST.

Spring Dean's List at KAIST.

Awarded Korea Presidential Science Scholarship from KOSAF (2015-2019).

Received Full Tuition Scholarship from KAIST (2015-2019).