Breadcrumb

Two DAC 2021 papers accepted

Two DAC 2021 paper from VSCLAB have been accepted. 

The first paper, led by PhD student Shyuan Yu, proposes a new approximate multiple using improved stochastic computing (called counter-based SC) techniques and showed significant advantages over existing state of the art approximate adders on DNN network applications. Congratulations on Shuyuan. 

  • COSAIM: Counter-based Stochastic-behaving Approximate Integer Multiplier for Deep Neural Networks

The second paper, led by PhD student Wentian Jin, developed a new graph convolution networks (DCN) based machine learning method to solve the electromigration stress analysis and show order of magnitude fast than the existing numerical approaches. Congratulations on Wentian. 

  • EMGraph: Fast Electromigration Stress Assessment for Interconnect Trees Using Graph Convolution Networks