▣ 제목(Title) : Neural network compression for on-device inference
▣ 연사(Speaker) : 조민식 연구원(Apple)
▣ 일시(Date &Time) : 2022.5.4(수) 1pm ~
▣ 언어(Language) : 한국어(Korean)
▣ Zoom URL : https://postech-ac-kr.zoom.us/j/97754060329?pwd=TUlkdjlqVGV6U1BiZzl4aEZVL2Y3Zz09
• Zoom ID : 977 5406 0329 Passcode : 251664

Dr. Minsik Cho is a machine learning researcher at Apple. Before joining Apple, he was with IBM research working on Deep-Learning/Machine-Learning/BigData acceleration through HW/SW codesign, and Scalable System design for Large-scale Deep-Learning. He received a Ph.D. in ECE from UT-Austin in 2008, and a BS in EE from SNU in 1999.

Deep neural network (DNN) model compression for efficient on-device inference is becoming increasingly important to reduce memory requirements and keep user data on-device. To this end, we propose a novel differentiable k-means clustering layer (DKM) and its application to train-time weight clustering-based DNN model compression. DKM casts k-means clustering as an attention problem and enables joint optimization of the DNN parameters and clustering centroids. Unlike prior works that rely on additional regularizers and parameters, DKM-based compression keeps the original loss function and model architecture fixed. We evaluated DKM-based compression on various DNN models for computer vision and natural language processing (NLP) tasks. Our results demonstrate that DKM delivers superior compression and accuracy trade-off on ImageNet1k and GLUE benchmarks. For example, DKM-based compression can offer 74.5% top-1 ImageNet1k accuracy on ResNet50 DNN model with 3.3MB model size (29.4x model compression factor). For MobileNet-v1, which is a challenging DNN to compress, DKM delivers 63.9% top-1 ImageNet1k accuracy with 0.72 MB model size (22.4x model compression factor). This result is 6.8% higher top-1accuracy and 33% relatively smaller model size than the current state-of-the-art DNN compression algorithms. Additionally, DKM enables compression of DistilBERT model by 11.8x with minimal (1.1%) accuracy loss on GLUE NLP benchmarks.