가상머신 및 최적화 연구실

Account-based blockchains such as Ethereum suffer from severe scalability and synchronization overhead due to continuously growing state data. Blockchain systems require strong guarantees for security, privacy, and trustworthiness in both on-chain and off-chain environments. Federated learning enables model training using edge devices while protecting client data privacy. However, there remain several limitations for practical deployment. Artificial neural networks tend to learn biased features when the training data itself contains bias. This research aims to explain why neural networks learn such biases and to develop algorithms that mitigate this issue. In particular, this study proposes the Neural Tangent Kernel (NTK) as an analysis framework and reveals why biased features are learned faster than other features. AI4Code focuses on advancing language models for code-related tasks, with the goal of achieving more effective and safer code generation. This research explores how different types of language models can be adapted to the code domain, particularly by improving their ability to generate correct, reliable, and secure code. A major focus of this research is on Diffusion Language Models, which provide an alternative generation paradigm to conventional autoregressive models. Soo-Mook Moon leads the Virtual Machine and Optimization (VM&O) Laboratory in the Department of Electrical and Computer Engineering at Seoul National University. VM&O Lab. is researching blockchains, compiler and platform optimizations, and a few AI topics.