数据挖掘20位学术新秀, 16名华人入选, 科大亮丽

在现今如火如荼的大数据时代，从学术研究到工业应用，从科学发现到医疗卫生服务，数据挖掘吸引了来自机器学习、统计、数据库、万维网、生物信息学、多媒体、自然语言处理、人机交互、社会网络计算、以及高性能计算等众多领域的专家和学者。ACM SIGKDD（简称KDD）作为数据挖掘领域的顶级国际会议，其评选出的20位KDD新星，也格外引人注目。

KDD论文录取率极低，百度表现不俗，但中国高校鲜有入选。不过20位KDD新星，却以华人学者为主，更有6位科大校友

近日，数据挖掘领域的顶级国际会议ACMSIGKDD （KDD2016）在美国旧金山召开。此次KDD大会堪称史上最大规模，有超过2700名来自学术界和工业界人士参与此次盛会，共吸引了1115篇投稿。其中研究专题投稿论文784篇，最终仅有142篇录用；应用数据科学专题投稿论文331篇，录用仅有66篇。可见大会论文竞争之激烈，水平之高超。此外，由于KDD的应用数据科学专题放开了过去只接收工业界投稿的限制，吸引了大量来自高校的学者投稿，因此今年的331篇投稿较之去年的189篇投稿有了大幅度的提升。

从以下录取论文的单位分布来看，Linkedin独领风骚，微软紧随其后，IBM和百度也表现不俗，显示工业界的强大实力。而领先的高校包括芝加哥大学、密歇根大学、和卡内基梅隆大学。国内高校没有一家入选两篇论文。

基于KDD大会的作者及论文等相关大数据信息，微软学术搜索评选出近6年里上升最快的二十位KDD学术新星。虽然国内研究单位在KDD入选论文方面尚有差距，引领数据挖掘潮流的学术新星，却有16位华人青年学者，其中更有6位科大校友，显示中国人在这一领域的十足后劲。下面是这些入选新星的介绍，和他（她）们的工作：

孙欢，加州大学圣塔芭芭拉分校  2010年毕业于中国科学技术大学

相关工作：

Exploiting RelevanceFeedback in Knowledge Graph Search;

Analyzing ExpertBehaviors in Collaborative Networks;

Network Mining andAnalysis for Social Applications;

Synthetic ReviewSpamming and Defense

Mohammad TahaBahadori   University of SouthernCalifornia

相关工作：

Multi-layerRepresentation Learning for Medical Concepts;

FLASH: Fast BayesianOptimization for Data Analytic Pipelines;

Deep ComputationalPhenotyping;

FBLG: a simple andeffective approach for temporal dependence discovery from time series data

任翔，伊利诺伊大学香槟分校  2012年毕业于浙江大学 (本科)

相关工作：

Label Noise Reduction inEntity Typing by Heterogeneous Partial-Label Embedding;

ClusType: EffectiveEntity Recognition and Typing by Relation Phrase-Based Clustering;

Automatic EntityRecognition and Typing from Massive Text Corpora: A Phrase and Network MiningApproach;

ClusCite: effectivecitation recommendation by information network-based clustering

Bee-Chung Chen  1998年毕业于国立台湾大学 (本科)

相关工作：

Ranking UniversitiesBased on Career Outcomes of Graduates;

An Empirical Study onRecommendation with Multiple Types of Feedback;

GLMix: GeneralizedLinear Mixed Models For Large-Scale Response Prediction;

Personalizing LinkedInFeed;

Activity ranking inLinkedIn feed;

SpatiotemporalPersonalized Recommendation of Social Media Content

刘洪甫，东北大学  毕业于北京航空航天大学

相关工作：

Infinite Ensemble forImage Clustering;

Spectral EnsembleClustering;

SEA: a system for eventanalysis on chinese tweets

曾春秋，佛罗里达国际大学  2009年毕业于四川大学 (硕士)

相关工作：

Online Context-AwareRecommendation with Time Varying Multi-Armed Bandit;

Applying data miningtechniques to address critical process optimization needs in advancedmanufacturing;

FIU-Miner: a fast,integrated, and user-friendly system for data mining in distributed environment

付燕杰，罗格斯大学  2008年毕业于中国科学技术大学 (本科)

相关工作：

Days on Market:Measuring Liquidity in Real Estate Markets;

Real Estate Ranking viaMixed Land-use Latent Models;

Exploiting geographicdependencies for real estate appraisal: a mutual perspective of ranking andclustering;

Learning geographicalpreferences for point-of-interest recommendation

Alex Deng，微软  2006年毕业于浙江大学 (本科)

相关工作：

Data-Driven MetricDevelopment for Online Controlled Experiments: Seven Lessons Learned;

Seven rules of thumb forweb site experimenters;

Online controlledexperiments at large scale;

Trustworthy onlinecontrolled experiments: five puzzling outcomes explained

赵亮，弗吉尼亚理工大学  

相关工作：

Hierarchical IncompleteMulti-source Feature Learning for Spatiotemporal Event Forecasting;

Multi-Task Learning forSpatio-Temporal Event Forecasting;

STED: semi-supervisedtargeted-interest event detectionin in twitter

Evangelos E. Papalexakis  Carnegie Mellon University

相关工作：

Whither Social Networksfor Web Search?;

Good-enough brain model:challenges, algorithms and discoveries in multi-subject experiments;

GigaTensor: scalingtensor analysis up by 100 times - algorithms and discoveries

袁晶，微软  2012年毕业于中国科学技术学 (博士)

相关工作：

Contextual IntentTracking for Personal Assistants;

Collaborative KnowledgeBase Embedding for Recommender Systems;

Regularity andConformity: Location Prediction Using Heterogeneous Mobility Data

Yang Zhou Georgia Institute of Technology

相关工作：

IntegratingVertex-centric Clustering with Edge-centric Clustering for Meta Path GraphAnalysis;

Activity-edge centricmulti-label classification for mining heterogeneous information networks;

Social influence basedclustering of heterogeneous information networks;

Multimedia features forclick prediction of new ads in display advertising

陈稳霖，华盛顿大学圣路易斯分校  2011年毕业于中国科学技术大学 (本科)

相关工作：

CompressingConvolutional Neural Networks in the Frequency Domain;

Optimal ActionExtraction for Random Forests and Boosted Trees;

Fast flux discriminantfor large-scale sparse nonlinear classification;

Density-based logisticregression;

An integrated datamining approach to real-time clinical monitoring and deterioration warning

Shuo Xiang，亚利桑那州立大学  2010年毕业于南京航空航天大学 (硕士)

相关工作：

Simultaneous feature andfeature group selection through hard thresholding;

Multi-source learningwith block-wise missing data for Alzheimer's disease prediction;

Robust principalcomponent analysis via capped norms;

Optimal exact leastsquares rank minimization

祝恒书，百度  2014年毕业于中国科学技术大学 (博士)

相关工作：

Recruitment Market TrendAnalysis with Sequential Latent Variable Models;

Days on Market:Measuring Liquidity in Real Estate Markets;

Talent Circle Detectionin Job Transition Networks;

Point-of-InterestRecommendations: Learning Potential Check-ins from Friends;

Taxi Driving BehaviorAnalysis in Latent Vehicle-to-Vehicle Networks: A Social Influence Perspective;

Real Estate Ranking viaMixed Land-use Latent Models;

Discerning TacticalPatterns for Professional Soccer Teams: An Enhanced Topic Model with Applications;

A cost-effectiverecommender system for taxi drivers;

Mobile apprecommendations with security and privacy awareness

张伟楠，伦敦大学学院  2011年毕业于上海交通大学 (本科)

相关工作：

Bid-aware GradientDescent for Unbiased Learning with Censored Data in Display Advertising;

Statistical ArbitrageMining for Display Advertising;

Annotating Needles inthe Haystack without Looking: Product Information Extraction from Emails;

Optimal real-timebidding for display advertising;

Joint optimization ofbid and budget allocation in sponsored search

Paulo Shakarian  Arizona State University, Military Academy

相关工作：

Early Identification ofViolent Criminal Gang Members;

Mining for CausalRelationships: A Data-Driven Study of the Islamic State;

Reducing gang violencethrough network influence based targeting of social programs;

Mining forgeographically disperse communities in social networks by leveraging distancemodularity

刘传仁，罗格斯大学  2008年毕业于中国科学技术大学 (本科)

相关工作：

Catch Me If You Can:Detecting Pickpocket Suspects from Large-Scale Transit Records;

UnifiedPoint-of-Interest Recommendation with Temporal Interval Assessment;

Data-driven AutomaticTreatment Regimen Development and Recommendation;

Temporal Phenotyping fromLongitudinal Electronic Health Records: A Graph Based Framework;

Temporal skeletonizationon sequential data: patterns, categorization, and visualization;

Proactive workflowmodeling by stochastic processes with application to healthcare operation and management;

A taxi businessintelligence system

Yasuko Matsubara  Kyoto University, Kumamoto University

相关工作：

Regime Shifts inStreams: Real-time Forecasting of Co-evolving Time Sequences;

FUNNEL: automatic miningof spatially coevolving epidemics;

Rise and fall patternsof information diffusion: model and implications;

Fast mining andforecasting of complex time-stamped events

蒋朦，伊利诺伊大学香槟分校  2010年毕业于清华大学 (本科)

相关工作：

CatchTartan:Representing and Summarizing Dynamic Multicontextual Behaviors;

CatchSync: catchingsynchronized behavior in large directed graphs;

FEMA: flexibleevolutionary multi-faceted analysis for dynamic behavioral pattern discovery