講座名稱：Large-Scale Datastreams Surveillance via Pattern-Oriented-Sampling

講座時(shí)間：2020-09-10 10:00

講座人：鄒長亮 教授

講座地點(diǎn)：騰訊會(huì)議直播（ID：230 606 623）

講座人介紹：

鄒長亮，南開大學(xué)統(tǒng)計(jì)與數(shù)據(jù)科學(xué)學(xué)院教授，副院長。2008年畢業(yè)于南開大學(xué)獲博士學(xué)位，隨后留校任教。主要從事統(tǒng)計(jì)學(xué)及其與數(shù)據(jù)科學(xué)領(lǐng)域的交叉研究和實(shí)際應(yīng)用。研究興趣包括：高維數(shù)據(jù)統(tǒng)計(jì)推斷、大規(guī)模數(shù)據(jù)流分析、變點(diǎn)和異常點(diǎn)檢測(cè)等，在Ann. Stat.、Biometrika、J. Am. Stat. Asso.、Math. Program.、Technometrics、IISE Tran.等統(tǒng)計(jì)學(xué)和工業(yè)工程領(lǐng)域權(quán)威期刊上發(fā)表論文幾十篇。

講座內(nèi)容：

Monitoring large-scale datastreams with limited resources has become increasingly important for real-time detection of abnormal activities in many applications. Despite the availability of large datasets, the challenges associated with designing an efficient change-detection when clustering or spatial pattern exists are not yet well addressed. In this talk, I will introduce a design-adaptive testing procedure when only a limited number of streaming observations can be accessed at each time. We derive an optimal sampling strategy, the pattern-oriented-sampling, with which the proposed test possesses asymptotically and locally best power under alternatives. Then, a sequential change-detection procedure is proposed by integrating this test with generalized likelihood ratio approach. Benefiting from dynamically estimating the optimal sampling design, the proposed procedure can improve the sensitivity in detecting clustered changes compared with existing procedures. Its advantages are demonstrated in numerical simulations and a real data example. Ignoring the neighboring information of spatially structured data will tend to diminish the detection effectiveness of traditional detection procedures.

主辦單位：數(shù)學(xué)與統(tǒng)計(jì)學(xué)院