Keynote Speakers 按姓氏拼音排列

Speaker

高新波

西安电子科技大学

Invited Speakers 按姓氏拼音排列

Speaker

纪荣嵘

厦门大学

Speaker

Hongdong Li

澳大利亚国立大学

Speaker

欧阳万里

香港中文大学

Speaker

吴飞

浙江大学

Speaker

吴建鑫

南京大学

Speaker

朱军

清华大学

讲者介绍

  高新波 个人主页 西安电子科技大学

报告题目:异质图像合成与识别

报告摘要:本次报告将以异质人脸图像合成和识别为例探讨人机混合智能和跨媒体智能的相关问题和思考。报告将系统总结我们团队近十年来在异质人脸图像识别,即素描画像和灰度图像之间的跨媒体识别工作,包括基于异质图像合成的方法和直接跨媒体识别方法,基于概率图模型的方法和基于深度学习的方法等。这些方法可以推广到其他异质图像的变换和识别中,如近红外与可见光图像,低分辨与高分辨图像,甚至CT和MR图像中,在在刑侦破案、网络追逃、动漫设计以及医学诊断等领域具有广泛的应用前景。

讲者信息:高新波,博士,教授,西安电子科技大学模式识别与智能系统学科带头人,综合业务网理论及关键技术国家重点实验室主任,国家万人计划科技创新领军人才,新世纪百千万人才工程国家级人选,国家杰出青年科学基金获得者,教育部长江学者特聘教授,科技部重点领域创新团队负责人、教育部创新团队负责人。IET Fellow、CIE Fellow、IEEE高级会员、中国图象图形学学会常务理事、中国计算机学会理事、中国指挥与控制学会富媒体指挥专委会常务委员、中国电子学会青年科学家俱乐部副主席。主要从事计算机视觉机器学习等领域的研究和教学工作,获国家自然科学二等奖1项、省部级科学技术一等奖3项。

  纪荣嵘 个人主页 厦门大学

报告题目:紧致化视觉大数据分析系统

报告摘要:报告主要探索视觉大数据搜索识别系统中的紧凑性问题,将覆盖纪荣嵘教授研究组近两年来在面向视觉终端应用的视觉特征紧凑表示和深度网络压缩中所做的一些工作与成果。在视觉特征紧凑表示方面,将介绍通过引入大规模无监督排序信息,学习排序敏感的哈希码,以保持原始高维特征空间中的检索信息。在深度网络压缩方面,将介绍面向特定任务(人脸和视觉场景解析)的深度网络级联压缩模型(串行低秩矩阵分解技术)与加速模型(结构化稀疏约束剪枝技术)。报告并将介绍上述研究在腾讯\滴滴\华为等视觉产品中的实际应用。

讲者信息:纪荣嵘,福建省“闽江学者”特聘教授,厦门大学教授、博士生导师、2014年获国家优青,2016年获国家万人计划青年拔尖。主要研究方向为计算机视觉与多媒体技术。相关工作发表于SCI源期刊论文90余篇,包括ACM汇刊与IEEE汇刊近50篇、CCF A类国际会议长文40余篇。论文的Google Scholar引用次数近5000次,SCI引用1600余次,H-因子为33,12篇论文入选ESI高被引/热点论文;近年来主持国家自然科学基金联合重点项目、军委科技委战略前沿专项,国家重点研发计划课题/子课题等;获2007年微软学者奖、2011年ACM Multimedia最佳论文奖、2012年哈工大优秀博士论文、2015年省自然科学二等奖、2016年教育部技术发明一等奖。担任多个国际期刊的副编辑,VALSE 2017大会主席、ACM/IEEE高级会员。

  Hongdong Li 个人主页 澳大利亚国立大学

报告题目:Some Recent Work on Non-Rigid Shape Structure-From-Motion with a Monocular Perspective Camera: Sparse and Dense Solutions.

报告摘要:In this talk, I will describe some of our recent work on monocular perspective camera based reconstruction of the non-rigid 3D shape or an object or a complex scene. We aim to answer an open question in 3D vision: “is it possible to recover the 3D shape of a dynamic deformable object with a single moving camera?”. Traditional methods for dynamic 3D reconstruction often employ stereo-vision, or assume the scene (with deformable object) follows certain simple low-order linear model. Our new work removes such restrictions and shows that, under certain mild assumptions, monocular 3D reconstruction of a dynamic shape scene is possible. I will explain two approaches, one is for the recovery of 3D dynamic human pose using structured movement, and the other is a dense surface reconstruction method for complex dynamic scene. Both methods achieved superior performance on standard benchmarks datasets.

讲者信息:Dr. Hongdong Li is an Associate Professor/Reader with the Computer Vision Group of ANU (Australian National University) and Chief Investigator for Australian Centre for Robotic Vision (ACRV). His research interests include 3D computer vision, SFM/SLAM for robot navigation and autonomous vehicles, as well as the application of mathematical optimization in geometric vision. He graduated from Zhejiang University, and taught at the same university before joined the ANU as a research fellow since 2004. During 2009-2010 he was a senior researcher with NICTA (Canberra Labs) working on the “Australia Bionic Eyes” project. He was a visiting professor with Carnegie Mellon University in 2017. He served as the Area Chair for CVPR, ICCV, ECCV, BMVC and 3DV in the past; Associate Editor for IEEE Transactions on PAMI (T-PAMI); Program Co-Chair for ACCV 2018. Jointly with students and co-workers he won a number of prestigious awards in computer vision, which include the CVPR Best Paper Award and Marr Prize-Honorable Mention.

  欧阳万里 个人主页 香港中文大学

报告题目:Exploring Deep Structures in Computer Vision tasks

报告摘要:Structure in data provide rich information that helps to reduce the complexity and improves the effectiveness of a model. In this talk, an introduction will be given on the recent progress in using deep learning as a tool for modeling the structure in visual data. We show that observation in our problem are useful in modeling the structure of deep model and help to improve the effectiveness of deep models for many vision problems.

讲者信息:Wanli Ouyang received the PhD degree in the Department of Electronic Engineering, The Chinese University of Hong Kong. He is now a senior lecturer at the University of Sydney. His research interests include image processing, computer vision and pattern recognition. He is the first author of 7 papers on TPAMI and IJCV, and has published around 40 papers on top tier conferences like CVPR, ICCV and NIPS. ImageNet Large Scale Visual Recognition Challenge (ILSVRC) is one of the most important grand challenges in computer vision. The team led by him ranks No. 1 in the ILSVRC 2015 and ILSVRC 2016. He receives the best reviewer award of ICCV. He has been the reviewer of many top journals and conferences such as IEEE TPAMI, TIP, IJCV, SIGGRAPH, CVPR, and ICCV. He is a senior member of the IEEE.

  吴飞 个人主页 浙江大学

报告题目:Memory-augmented learning

报告摘要:Neural networks with a memory capacity provide a promising approach to media understanding (e.g., Q-A and visual classification). In this talk, I will present how to utilize the information in external memory to boost media understanding. In general, the relevant information (e.g., knowledge instance and exemplar data) w.r.t the input data is sparked from external memory in the manner of memory-augmented learning. Memory-augmented learning is an appropriate method to integrate data-driven learning, knowledge-guided inference and experience exploration.

讲者信息:Fei Wu is a professor at the college of computer science, Zhejiang University. From October, 2009 to August 2010, Fei Wu was a visiting scholar at Prof. Bin Yu's group, University of California, Berkeley. Currently, he is the vice-dean of college of Computer Science, and the director of Institute of Artificial Intelligence of Zhejiang University. His research interests mainly include Artificial Intelligence, cross-media computing, and multimedia retrieval. He has won various honors such as the Award of National Science Fund for Distinguished Young Scholars of China (2016).

  吴建鑫 个人主页 南京大学

报告题目:深度学习实践:庖丁解牛与盲人摸象

报告摘要:深度学习是目前计算机视觉与机器学习领域最热门、也是在很多实际问题中实践效果最好的方法。然而,深度学习,尤其是卷积神经网络CNN的机理目前尚不明确。本次报告将介绍我们研究组在CNN深度学习方向上两个方面的实践:庖丁解牛与盲人摸象。庖丁解牛即将CNN的各个构成模块分别探索,发现其优缺点并加以改进,从而对CNN的各个模块形成深入的理解。盲人摸象即对在ImageNet上学习到的CNN预训练模型能起到什么样的作用加以研究,在CNN整体机理尚不清楚的前提下,对预训练模型在各个视觉问题中的无监督应用加以研究。

讲者信息:南京大学教授,Minieye首席科学家(minieye.cc)。研究兴趣为计算机视觉与机器学习,尤其是资源(计算、存储、能源、数据与标注)受限情况下的深度学习。曾获中组部青年千人和基金委优秀青年科学基金资助,曾任ICCV、CVPR等领域主席。

  朱军 个人主页 清华大学

报告题目:

报告摘要:深度生成模型是一类灵活的从复杂数据中提取隐含结构并且进行“从上到下”生成样本的方法,广泛用于无监督学习、半监督学习等任务。该报告将介绍最近的一些前沿进展,包括用于半监督学习的模型和算法,以及支持快速编程实现的珠算(ZhuSuan)编程库。

讲者信息:朱军,清华大学长聘副教授、卡内基梅隆大学兼职教授、智能技术与系统国家重点实验室副主任。2001到2009年获清华大学计算机学士和博士学位,之后在卡内基梅隆大学做博士后,2011年回清华任教。主要从事人工智能基础理论、高效算法及相关应用研究,在国际重要期刊与会议发表学术论文近百篇。受邀担任人工智能顶级杂志IEEE TPAMI和AI的编委、《自动化学报》编委,担任机器学习国际大会ICML2014地区联合主席, ICML (2014-2018)、NIPS (2013, 2015)、UAI (2014-2017)、IJCAI(2015,2017)、AAAI(2016-2018)等国际会议的领域主席,中国计算机学会(CCF)学术工委主任助理。获CCF优秀博士论文奖、CCF自然科学一等奖、CCF青年科学家奖、国家优秀青年基金、北京市优秀青年人才奖、清华大学优秀班主任一等奖等,入选国家“万人计划”青年拔尖人才、IEEE Intelligent Systems杂志评选的“AI’s 10 to Watch”及清华大学221基础研究人才计划。指导的学生获NIPS 2017国际人工智能对抗攻防竞赛全部三个任务的冠军。