Jin-Duk Park

Ph.D. candidate, dept. of CSE,
Yonsei Univ, Seoul, Korea.
jindeok6@yonsei.ac.kr

(MIDaS lab,
advisor: Prof. Won-Yong Shin)

Talk Material (Selected)

- KAN: Kolmogorov–Arnold Networks

Note : KAN; MLP; KA representation theorem
[Presentation file]

- Mixture of Experts

Note : MoEs; LLM; GPUs; LLM inference
[Presentation file]

- Graph Filtering for Collaborative Filtering

Note : CF; graph filtering; GCN
[Presentation file] [Record link]

- Question Answering

Note : natural language; question answering
[Presentation file]

- Joint Visual & Text Learning

Note : Multi-modal; visual learning; natural language; text learning
[Presentation file]

- Graph Signal Processing and Graph Filter Design

Note : graph signal processing; graph filtering;
[Presentation file]

- Physics-Informed Graph Neural Network Methods

Note : PINN, equivariant GNNs
[Presentation file]

- Graph Neural Networks for Heterogeneous Graph

Note : HAN, GTN, HG embdding
[Presentation file]

- Toward Explainable AI (XAI): Deep Taylor Decomposition

Note : DTD, XAI
[Presentation file]

- Initialization Techniques of Neural Network Parameters

Note : Xavier intialization, He initialization
[Presentation file]

- Introduction to Recommender System

Note : collaborative filtering, cold start, matrix factorization
[Presentation file]