Hello, I'm Dongseok Kim, a Computer Scientist specializing in AI/ML with a focus on general-purpose methodologies and their theoretical foundations. My work isn’t tied to any single domain — I’m interested in principles that cut across fields and can be applied broadly in meaningful ways.

I see my research as sitting between theory and practice. Mathematics and statistics are not just tools for me; they provide the grounding needed to make AI systems stable and trustworthy. At the same time, I rely on experiments and intuition to spark new ideas, aiming to design methods that are both rigorous and useful in real-world settings.

I take an interdisciplinary perspective, drawing on concepts from different academic traditions to push AI research in new directions. I’m less interested in narrow techniques or fixed applications and more motivated by combining diverse approaches to build frameworks that are robust, sustainable, and widely usable.

I’m also more drawn to exploring uncharted territory than following well-established paths. Ensuring validity through careful analysis matters to me, but so does keeping my work connected to practical impact. In the long run, I don’t want to be a specialist confined to one corner of the field. Instead, I aim to bridge disciplines and contribute to shaping new paradigms in AI, creating foundations that others can extend and apply across a wide range of contexts.

DongSeok Kim, Shabir Ahmad, TaegKeun Whangbo
Federated Regressive Learning: Adaptive Weight Updates through Statistical Information of Clients
Applied Soft Computing, 2024

Wonjun Jeong, Dongseok Kim, Taegkeun Whangbo
SCOPE: Stochastic and Counterbiased Option Placement for Evaluating Large Language Models
arXiv Preprint, 2025

Dongseok Kim, Wonjun Jeong, Gisung Oh
Convergence and Generalization of Anti-regularization for Parametric Models
arXiv Preprint, 2025

Dongseok Kim, Hyoungsun Choi, Mohamed Jismy Aashik Rasool, Gisung Oh
Gaming and Cooperation in Federated Learning: What Can Happen and How to Monitor It
arXiv Preprint, 2025

Dongseok Kim, Hyoungsun Choi, Mohamed Jismy Aashik Rasool, Gisung Oh
ORACLE: Explaining Feature Interactions in Neural Networks with ANOVA
arXiv Preprint, 2025

Dongseok Kim, Hyoungsun Choi, Mohamed Jismy Aashik Rasool, Gisung Oh
CLAPS: Posterior-Aware Conformal Intervals via Last-Layer Laplace
arXiv Preprint, 2025

Dongseok Kim, Hyoungsun Choi, Mohamed Jismy Aashik Rasool, Gisung Oh
$\phi$-test: Global Feature Selection and Inference for Shapley Additive Explanations
arXiv Preprint, 2025

Dongseok Kim, Hyoungsun Choi, Mohamed Jismy Aashik Rasool, Gisung Oh
Theoretical Foundations of Prompt Engineering: From Heuristics to Expressivity
arXiv Preprint, 2025