AI Computing Platform Laboratory
Department of Computer Science and Engineering, Artificial Intelligence Convergence, Ewha Womans University
The AI Computing Platform Laboratory (ACPL) has been a lab of excellence for efficient AI hardware & software research since its founding in 2021.
Our main research goal is to accelerate AI model in a faster and energy-efficient way through HW/SW co-design. Specifically, our research interests include:
- Designing Neural Processing Unit (NPU) and domain-specific hardware
- Making AI model efficient, such as quantization, pruning, and knowledge distillation
- Hardware-aware neural architecture search (HW-Aware NAS) and neural architecture accelerator search (NAAS)
- Processing-in-memory (PIM)
news
Oct 11, 2024 | Doeun Kim has joined our group as an undergraduate intern. Welcome! |
---|---|
Oct 01, 2024 | Suhyeon Kim has joined our group as an undergraduate intern. Welcome! |
Sep 19, 2024 | Jaeyoung Choi has joined our group as an undergraduate intern. Welcome! |
Sep 14, 2024 | Our lab has two papers accepted for presentation at 2024 International Conference on Communications, Computing, Cybersecurity, and Informatics (CCCI). |
Sep 05, 2024 | Wonhui Roh and Jungyeon Ha have joined our group as an undergraduate intern. Welcome! |
latest publications
-
AutoCaps-Zero: Searching for Hardware-Efficient Squash Function in Capsule NetworksIn 2024 International Conference on Communications, Computing, Cybersecurity, and Informatics (CCCI)
-
OCW: Enhancing Few-Shot Learning with Optimized Class-Weighting MethodsIn 2024 International Conference on Communications, Computing, Cybersecurity, and Informatics (CCCI)
-
An Energy-Efficient Hardware Accelerator for On-Device Inference of YOLOXIn 2024 21st International SoC Design Conference (ISOCC)
-
BS2: Bit-Serial Architecture Exploiting Weight Bit Sparsity for Efficient Deep Learning AccelerationIn 2024 21st International SoC Design Conference (ISOCC)
-
AlphaAccelerator: An Automatic Neural FPGA Accelerator Design Framework Based on GNNsIn 2024 21st International SoC Design Conference (ISOCC)