About

Representation Analysis & Application Lab. @ Inha University

We study how and why deep learning models succeed or fail in real-world prediction problems, with a particular focus on understanding and analyzing learned representations.

Our research is grounded in information-theoretic principles, which we use as quantitative tools to analyze generalization, diagnose representation collapse, and assess the reliability of neural networks.

Building on this foundation, we develop machine learning algorithms for practical and deployable prediction systems, especially in settings involving tabular and structured data, such as finance, healthcare, and industrial applications.

Our current research interests include:

• Information-theoretic analysis and diagnosis of generalization in deep neural networks
• Reliable and deployable learning and inference on tabular data
• Multimodal learning with structured data for real-world decision systems

📌 Prospective Students
Students interested in joining the lab as undergraduate researchers or MS/Ph.D. students may contact me via email. (학부 연구생 또는 석·박사 과정 진학에 관심 있는 학생은 메일로 연락 바랍니다.)