報(bào)告題目：非中心化聯(lián)邦學(xué)習(xí)的統(tǒng)計(jì)推斷

報(bào)告人：陳松蹊 中國科學(xué)院院士 北京大學(xué)

主持人：鄭志明 中國科學(xué)院院士 北京航空航天大學(xué)

報(bào)告時(shí)間：2024年9月8日（星期日）11:00

報(bào)告地點(diǎn)：湘潭大學(xué)研究生院報(bào)告廳

報(bào)告摘要：

This paper considers decentralized Federated Learning (FL) under heterogeneous distributions among distributed clients or data blocks for the M-estimation. The mean squared error and consensus error across the estimators from different clients via the decentralized stochastic gradient descent

algorithm are derived. The asymptotic normality of the Polyak-Ruppert (PR) averaged estimator in the decentralized distributed setting is attained, which shows that its statistical efficiency comes at a cost as it is more restrictive on the number of clients than that in the distributed M-estimation. To overcome the restriction, a one-step estimator is proposed which permits a much larger number of clients while still achieving the same efficiency as the original PR-averaged estimator in the non-distributed setting. The confidence regions based on both the PR-averaged estimator and the proposed one-step estimator are constructed to facilitate statistical inference for decentralized Federated Learning.

報(bào)告人簡介：

陳松蹊，中國科學(xué)院院士，全國政協(xié)委員，北京大學(xué)講席教授，北京大學(xué)數(shù)學(xué)科學(xué)學(xué)院和光華管理學(xué)院教授，中國概率統(tǒng)計(jì)學(xué)會(huì)理事長。主要研究方向?yàn)槌呔S大數(shù)據(jù)統(tǒng)計(jì)分析、環(huán)境統(tǒng)計(jì)、非參數(shù)統(tǒng)計(jì)方法等，在超高維假設(shè)檢驗(yàn)方法和非參數(shù)經(jīng)驗(yàn)似然方法方面取得豐碩成果；注重?cái)?shù)理統(tǒng)計(jì)的應(yīng)用，以國家大氣污染防治的重大需求為出發(fā)點(diǎn)，從事統(tǒng)計(jì)學(xué)與大氣環(huán)境交叉學(xué)科研究，提出了去除大氣監(jiān)測(cè)數(shù)據(jù)中的氣象因素干擾的方法，為精準(zhǔn)度量污染排放和評(píng)估大氣治理效果提供了科學(xué)方法。曾獲教育部自然科學(xué)一等獎(jiǎng)，被選為美國科學(xué)促進(jìn)會(huì)會(huì)士、美國統(tǒng)計(jì)學(xué)會(huì)會(huì)士、數(shù)理統(tǒng)計(jì)學(xué)會(huì)會(huì)士。

湖南國家應(yīng)用數(shù)學(xué)中心

湖南韶峰應(yīng)用數(shù)學(xué)研究院

到家集團(tuán)

湘潭大學(xué)數(shù)學(xué)與計(jì)算科學(xué)學(xué)院

湖南省數(shù)學(xué)學(xué)會(huì)

湘潭市國家高新技術(shù)產(chǎn)業(yè)開發(fā)區(qū)

湘潭市大數(shù)據(jù)和產(chǎn)業(yè)創(chuàng)新發(fā)展中心

“智能計(jì)算與信息處理”教育部重點(diǎn)實(shí)驗(yàn)室

“科學(xué)工程計(jì)算與數(shù)值仿真”湖南省重點(diǎn)實(shí)驗(yàn)室

“計(jì)算科學(xué)”湖南省科技創(chuàng)新國際合作基地