School of Computer Science and Technology,
Guangdong University of Technology, China
Zhenguo Yang is an Associate Professor at Guangdong University of Technology. He was a Postdoctoral Fellow at Guangdong University of Technology from 2017 to 2019. He received his Ph.D. degree in Computer Science from City University of Hong Kong, in 2017. His research interests include event detection, domain adaptation cross-modal retrieval, etc. He has published over 40 publications, including TPAMI, ACM TOIT, ACM MM, WWWJ, ICMR, MMM, PCM, etc. He has got Top Performance Award of the Grand Challenges on ACM MM 2017/2018/2019.
Cross-domain and Cross-modality Event Mining and Retrieval
Online news media and social media provide amounts of multimedia resources that are publicly available. The data records people’s daily lives, which are related to real-world events, e.g., protests, celebrations, natural disasters, festivals, politics, etc. This tutorial focuses on how to organizing the multi-domain and multi-modality data distributed on the Internet platforms according to their depicted real-world events. In this tutorial, event discovery and retrieval tasks crossing data domains and modalities will be introduced. The challenging issues will be discussed, and a few approaches from the perspectives of matrix factorization and deep learning will be presented, followed by the analyses on the experimental results. In particular, a few multi-domain and multi-modality real-world event datasets collected by the speakers will be introduced, which have been released on GitHub for peer researchers (https://github.com/zhengyang5).
Department of Electrical Engineering,
National Taiwan University, Taiwan
Hung-yi Lee received the M.S. and Ph.D. degrees from National Taiwan University (NTU), Taipei, Taiwan, in 2010 and 2012, respectively. From September 2012 to August 2013, he was a postdoctoral fellow in Research Center for Information Technology Innovation, Academia Sinica. From September 2013 to July 2014, he was a visiting scientist at the Spoken Language Systems Group of MIT Computer Science and Artificial Intelligence Laboratory (CSAIL). He is currently an associate professor of the Department of Electrical Engineering of National Taiwan University, with a joint appointment at the Department of Computer Science & Information Engineering of the university. His research focuses on machine learning (especially deep learning), spoken language understanding and speech recognition. He owns a YouTube channel teaching deep learning (in Mandarin) with more than 3.5M views and 44k subscribers.
Generative Adversarial Network and its Applications to Natural Language and Speech Processing
Generative adversarial network (GAN) is a new idea for training models, in which a generator and a discriminator compete against each other to improve the generation quality. Recently, GAN has shown amazing results in image generation, but the applications of GAN on text and speech processing are still limited. In this talk, I will demonstrate the applications of GAN on voice conversion, unsupervised speech recognition, unsupervised abstractive summarization, and sentence generation.
School of Software Engineering,
South China University of Technology, China
Yi Cai is a professor in South China University of Technology, and he received his PhD from the Chinese University of Hong Kong in 2009. Before he joined SCUT, he was a post-doctor of City University of Hong Kong. He was a visiting scholar of Imperial College London, Tsinghua University, City University of Hong Kong, Nanyang Technological University. He published more than 100 papers (e.g., IEEE Transactions on Knowledge and Data Engineering, Neural Networks, Decision Support Systems, INFORMS Journal on Computing, Knowledge Based Systems, AAAI, EMNLP, AAMAS, CIKM, ER and ICTAI) and 2 books. He reviewed papers from conferences and journals related to information retrieval, semantic web, recommender system, data mining and database, including TKDE, TOIT, Decision Support Systems, WWWJ, KBS, JCST, CIKM and ER. He is a program committee member of conferences, including ICWL, ICWE, WAIM, ICEBE and NDBC. He is the co-chair of Social Networking and Mining Track in EIDWT-2013, DaSem 2013, SeCop 2015 and APWeb-WAIM 2018.
Leveraging Knowledge for Short Text Analysis
Since some documents are short and ambiguous, recent research enriches document representation with concepts of words extracted from an external knowledge base. However, this approach might incorporate task-irrelevant concepts or coarse granularity concepts that could not discriminate classes in a text classification task. This might bring noise to document representation and degrade text analysis performance. In this talk, we will introduce the latest developments of leveraging knowledge into short text analysis.
School of Data and Computer Science,
Sun Yat-sen University, China
Yanghui Rao is an Associate Professor with the School of Data and Computer Science, Sun Yat-sen University, Guangzhou, China. He received the Ph.D. degree from the City University of Hong Kong in 2014, and the master's degree from the Graduate University of the Chinese Academy of Science in 2010. His current research interests include emotion detection, sentiment analysis, topic modeling, and natural language processing. He has published over 20 refereed journal and conference papers, including the ACM Transactions on Information Systems, the IEEE Transactions on Cybernetics, the IEEE Transactions on Affective Computing, IEEE Intelligent Systems, ACL, EMNLP, CIKM, and DASFAA.
Sentiment and Emotion Detection over Text
The Web has generated product reviews, sentimental and emotional text about brand perception continuously, which can be utilized for capturing the opinions of consumers and the general public on product preferences, company strategies, and marketing campaigns. In this tutorial, the following topics will be discussed. The first topic is the detection of sentiments and emotions from short messages which is prevalent in social networks. The second topic is how to deal with noisy labels when conducting sentiment and emotion classification, because the ground truth for sentiments and emotions in social media is often constructed through surveys, hashtags or emoticons, where the labels may contain many errors. Last but not the least, sentiment and emotion detection over cross-domain corpora will be explored.