김재준 연구실

Model Compression including Neural Network Pruning and Quantization, and Efficient Serving Systems for Generative Models (Diffusion, LLM) and LLM Agent and Multi-Modal Systems

Variable-Precision Neural Network Hardware, Quantization-Aware Neural Network Hardware, Efficient Point Cloud Acceleration, and Hardware-Network Co-Design for LLM

Digital Computing Logic near/in-memory including Software/Hardware co-design for DRAM-PIM and Logic and MAC Operations in Flash Memory, Analog Compute-in-Memory (ACIM) Design, and Digital Compute-in-Memory (DCIM) Design

Technology-Circuit Co-Design for Emerging Memory including Process Architecture Design and Modeling/Circuit Design for Emerging Technologies, and Circuit Design and Optimization for 3D Memory