报告人：毛先领

简介：毛先领，北京理工大学副教授，博导。主要研究深度学习、机器学习与网络数据挖掘，具体研究Information Extraction、 Question Answering and Dialogue和Learn to Hashing等方向。目前担任计算机学会中文信息技术专委会委员，中文信息学会青工委委员以及语言与知识专委会委员；已在SIGIR、AAAI，IJCAI, TOIS, TKDE, CIKM, EMNLP, COLING等国际期刊会议上发表30余篇论文；分别获NLPCC 2019和ICKG 2020最佳论文奖；部分成果获中国电子学会科技进步一等奖（2018）和浙江省科技进步三等奖（2018）；正在承担或参与国家重点研发计划子课题、国家自然科学基金重点项目和面上项目等多项。

题目:  Similarity-preserved Hashing: Diffusing from Images Retrieval to Other Scenarios

报告时间：2021年11月25日9:00—11:00

报告地点：信息大厦B1111

Abstract:In the past decade, we have witnessed an explosive growth of data on the Internet, and it brings both challenges and opportunities to traditional algorithms developed on small to median scale data sets. Particularly, nearest neighbor search (NN) has become a key ingredient in many large-scale machine learning and data management tasks. In fact, approximate nearest neighbors (ANN) are enough to achieve satisfactory performance in many applications, such as the image retrieval task in search engines. Due to the low storage cost and fast retrieval speed, similarity-preserved hashing is one of the popular solutions for ANN search. This talk will first review related methods for images, then introduce the ways how similarity-preserved hashing is enabling natural language processing. It will also highlight open problems that are being addressed by emerging research.