Deep Learning for Natural Language Processing and Beyond

The Tutorial begins with the basic of feed-forward neural network and relevant fundamental knowledge for deep learning. We then introduce more specialized neural network models, including Convolutional Neural Network, Recurrent Neural Network, and attention-based models. In the second part, we will present how these models and techniques can be applied to some interesting problems of natural language processing including sentiment classification, textual entailment recognition, natural language generation, and question answering. The last part of the tutorial will show how we can adapt deep learning and natural language processing techniques for program analysis.

Assoc. Prof. Nguyen Le Minh – Japan Advanced Institute for Science and Technology (Japan)

Minh Le Nguyen is currently an Associate Professor of School of Information Science, JAIST. He leads the lab on Machine Learning and Natural language Understanding at JAIST. He received his B.Sc. degree in information technology from Hanoi University of Science, and M.Sc. degree in information technology from Vietnam National University, Hanoi in 1998 and 2001, respectively. He received his Ph.D. degree in Information Science from School of Information Science, Japan Advanced Institute of Science and Technology (JAIST) in 2004. He was an assistant professor at School of information science, JAIST from 2008-2013. His research interests include machine learning, natural language understanding, question answering, text summarization, machine translation, big data mining, and Deep Learning.

Social graph analysis for verification and fill out users’ information

Dữ liệu mạng xã hội bên cạnh dữ liệu hồ sơ người dùng còn có thêm thông tin về quan hệ của họ với những người khác. Hai nhóm thông tin này có mối quan hệ chặt chẽ với nhau. Do vậy, nếu được phân tích đúng chúng có thể được dùng để bổ sung hoặc kiểm tra chéo lẫn nhau. Một trong những khó khăn lớn nhất của việc ứng dụng dữ liệu mạng xã hội trong kinh doanh là dữ liệu hồ sơ người dùng rất thưa và kém chính xác. Bằng việc sử dụng một số thuật toán embedding thông tin về mối quan hệ xã hội và các thuật toán dự đoán, phân lớp khác trên đó, ta có thể kiểm chứng hoặc/và bổ sung thông tin hồ sơ của người dùng để chúng trở nên hoàn thiện và có ý nghĩa kinh doanh hơn.

Mr. Le Minh – Five9 (Vietnam)

AI in Electro-Optical/Infrared camera surveillance systems

Nowadays, Electro-Optical and Infrared (EO/IR) technology plays a critical role in many military, defense, security and industry applications; as it provides the day-night and long-range visualization capability, improves the user’s ability to automatically identify targets, performs threat assessment, raises situational awareness, as well as supports weapons engagement through automatic surveillance and fire control solutions through line-of-sight. In this talk, we will introduce several modern EO/IR camera systems that are currently being developed in Viettel R&D Institute. Furthermore, we will propose a number of artificial intelligence applications that equip our EO/IR camera surveillance systems with the ability to automatically perform detection, localization, recognition, identification and tracking of all ground/air/maritime target types in real time. Our solutions are based on some most advanced machine-learning and deep-learning models trained on large-scale data. Several techniques of online learning and multi-sensor fusion will also be proposed to provide our system with high-performance accuracies and low false alarm rates, even in bad-seeing conditions or complex backgrounds.

Dr. Dao Duc Minh – Viettel R&D (Vietnam)

Dr. Minh Dao received the B.Sc. degree in Electrical Engineering from Hanoi University of Technology, Vietnam in 2007, the double Master degree in Information and Communication Technologies from Polytechnic University of Turin, Italy, and Karlsruhe Institute of Technology, Germany in 2009, and the Ph.D.in Electrical and Computer Engineering from The Johns Hopkins University, Baltimore MD, USA, in 2015. From 2015 to 2017, he worked as a Research Scientist in the U.S. Army Research Laboratory (ARL) in Maryland, USA. In June of 2017, he joined Viettel R&D Institute, where he currently leads the Image Processing R&D team. His research interests are broadly in the areas of signal/image/video processing, statistical machine learning, computer vision and artificial intelligence.

Structured prediction for the summarisation and alignment of videos

In this talk, I present two approaches for the automated summarisation and the automated alignment of videos based on structured prediction. The first approach aims to summarise an action video by a selection of its frames while simultaneously recognising its main action. The second approach aims to find the best alignment of two given videos and can be used as a generalised distance for video classification. Both approaches are similar in spirit and leverage structured prediction and the structural SVM framework for inference and for training. Potentially, they could be coupled with deep learning layers to capture the best of both worlds.

Prof. Massimo Piccardi – University of Technology Sydney (Australia)

Massimo Piccardi (M.Eng., 1991, Ph.D. Bologna, 1995) is a professor at University of Technology Sydney (UTS) where he serves as a discipline leader for the School of Electrical and Data Engineering and as a program leader for the Global Big Data Technology Centre, a University-supported task force in communications and big data analytics. His main research areas are computer vision, multimedia, machine learning, and, more recently, natural language processing. Prof. Piccardi has published over a hundred and fifty papers to date, is a Senior Member of the IEEE and serves as an Associate Editor for journal “Computer Vision and Image Understanding.