報(bào)告題目:Structural Design for Efficient Deep Network Implementation
報(bào)告人:Yu Hen Hu(胡玉衡)
報(bào)告時(shí)間:2019年5月29日14:30-16:00
報(bào)告地點(diǎn):友誼校區(qū)航海學(xué)院東配樓
邀請(qǐng)人:趙瑞琴 副教授
報(bào)告摘要 :The Deep Neural Network (DNN) has significantly enhanced the performance of pattern classification and time series predictions for numerous important real-world applications. However, the complexity of a typical DNN network has also made it difficult to incorporate such an algorithm in low power mobile devices and internet of things. In this work, we present methods to approximate the performance of a trained DNN by modifying its structural design and discuss specific design cases.

報(bào)告人簡(jiǎn)介:美國(guó)University of Wisconsin-Madison電子與計(jì)算機(jī)工程系,教授,系主任,IEEE Fellow。他于1976年在國(guó)立臺(tái)灣大學(xué)獲得學(xué)士學(xué)位,分別于1980和1982年獲美國(guó)南加州大學(xué)碩士和博士學(xué)位。在1983到1987年間,他在Southern Methodist University電子工程系擔(dān)任助理教授,從1987年開(kāi)始到University of Wisconsin-Madison工作,目前為教授、電子與計(jì)算機(jī)工程系執(zhí)行主任。Dr. Hu研究興趣廣泛,主要包括:信號(hào)處理算法的設(shè)計(jì)和實(shí)現(xiàn)、大規(guī)模集成電路計(jì)算機(jī)輔助設(shè)計(jì)、模式分類(lèi)和機(jī)器學(xué)習(xí)算法、圖像信號(hào)處理等。他發(fā)表了超過(guò)200篇高水平學(xué)術(shù)文獻(xiàn),并出版過(guò)多部專(zhuān)著。他曾經(jīng)擔(dān)任IEEE Transaction of Acoustic, Speech, and Signal Processing、IEEE signal processing letters等多個(gè)重要國(guó)際期刊的副編輯,還曾擔(dān)任IEEE信號(hào)處理協(xié)會(huì)執(zhí)行委員會(huì)成員及秘書(shū)、IEEE信號(hào)處理協(xié)會(huì)神經(jīng)網(wǎng)絡(luò)委員會(huì)和多媒體信號(hào)處理委員會(huì)主席。他還是Multimedia and Expo國(guó)際會(huì)議和IEEE Transactions on Multimedia on behalf of IEEE Signal processing society 咨詢(xún)委員會(huì)成員。