초청 세미나가 아래와 같이 12월 14일(화)에 진행되오니 관심 있는 분들의 많은 참석 부탁 드립니다.
- 아 래 -
연사: 강석주 교수 (서강대학교 전자공학과)
제목: 딥러닝 모델 경량화 및 하드웨어 가속 기술 동향
일시: 2021년 12월 14일(화) 오후 3시
장소: Zoom 온라인 (비대면)
온라인 참여 희망자: hanjaeho@korea.ac.kr로 이메일 요청 시 zoom 링크 발송 예정
초록:
In this seminar, lightweight-based deep learning technology and hardware accelerator technology will be explained. In addition, various existing methodologies and two proposed methods in our laboratory will be explained. In the first topic, generative adversarial networks (GANs) have shown excellent performance in image and speech applications and create impressive data primarily through a new type of operator called deconvolution (DeConv) or transposed convolution (Conv). To implement the DeConv layer in hardware, the state-of-the-art accelerator reduces the high computational complexity via the DeConv-to-Conv conversion and achieves the same results. However, there is a problem that the number of filters increases due to this conversion. Recently, Winograd minimal filtering has been recognized as an effective solution to improve the arithmetic complexity and resource efficiency of the Conv layer. Hence, we propose an efficient Winograd DeConv accelerator that combines these two orthogonal approaches on FPGAs. In the second topic, deformable convolutional networks have demonstrated outstanding performance in object recognition tasks with an effective feature extraction. Unlike standard convolution, the deformable convolution decides the receptive field size using dynamically generated offsets, which leads to an irregular memory access. Thus, a naive implementation would lead to an excessive memory footprint. Therefore, we present a novel approach to accelerate deformable convolution on FPGA using a novel training method to reduce the size of the receptive field in the deformable convolutional layer without compromising accuracy.