Speech Title:    Fuzzy Transfer Learning

Abstract:
This talk will describe how fuzzy transfer learning can innovatively and effectively learn from data to support data-driven decision-making in uncertain and dynamic situations. The core idea behind fuzzy transfer learning is to leverage previously acquired knowledge to assist in completing a prediction task in a related domain by integrating fuzzy techniques with the transfer learning process. A set of new fuzzy transfer learning theories, methodologies, and algorithms is introduced, which transfers knowledge learned in one or more source domains to target domains. The fuzzy transfer learning set incorporates (1) a fuzzy refinement domain adaptation algorithm by utilizing the fuzzy system and similarity/dissimilarity concepts to modify the target instances' labels for classification; (2) fuzzy rule-based systems with mapping functions by building latent spaces to facilitate knowledge transfer for regression tasks in both homogeneous and heterogeneous scenarios; (3) unsupervised domain adaptation, to recognize newly emerged patterns in target domains that may be unlabelled. Patterns in target domains are recognized by leveraging knowledge from patterns learned from source domains and solutions to heterogeneous unsupervised domain adaptation via n-dimensional fuzzy geometry and fuzzy equivalence relations. These new developments can enhance data-driven prediction and decision support systems in complex real-world environments.

Short Bio:
Distinguished Professor Jie Lu is a scientist in the field of computational intelligence, primarily known for her work in fuzzy transfer learning, concept drift, recommender systems, and decision support systems. She is an IEEE Fellow, IFSA Fellow, and Australian Laureate Fellow. Currently, Prof Lu is the Director of the Australian Artificial Intelligence Institute (AAII) and Associate Dean (Research Excellence) at the Faculty of Engineering and Information Technology, University of Technology Sydney (UTS). She has published over 400 papers in leading journals and conferences; won 10 Australian Research Council (ARC) Discovery Projects and led 15 industry projects; and supervised 50 doctoral students to completion. Prof Lu serves as Editor-In-Chief for Knowledge-Based Systems and International Journal of Computational Intelligence Systems, and is a recognized keynote speaker, delivering 30 keynote speeches at international conferences. She is the recipient of the IEEE Transactions on Fuzzy Systems Outstanding Paper Award (2019), the Computer Journal Wilkes Award (2018), Australia's Most Innovative Engineer Award (2019), and the UTS Chancellor's Medal for Research Excellence (2019).

Speech Title:    Empowering Cognitive Security Systems with Cyber Situation Awareness

Abstract:
To solve the pressing security challenges of our era related also to threats such as terrorism and cyberterrorism, we need more creative approaches able to detect connections between relations, events and concepts in evolving contexts characterized by structured and unstructured data coming up from multiple sensors and human based networks. In this talk I propose the adoption of Cyber Situation Awareness (SA) to define systems able to handle such issues. Endsley provided a formal definition that considers SA as a three-phase process: perception of the elements of an environment in each time interval, comprehension of these elements and projection of their states into the near future. In detail, from a computational viewpoint, Cyber SA aims at formalizing and deducing situations (occurring in the real world) by processing, fusing and abstracting data and information. The Cyber SA approach needs to be populated with computational intelligence techniques, such as Granular Computing, fuzzy and rough sets, able to gather, pre-process, aggregate and filter data coming from sensors (both physical - coming from sensing devices - and virtual - coming from social networks), to sustain conceptualization and reasoning on situations. In this talk I present some concrete examples of applying the Cyber SA to safety & security and Intelligence Analysis for counterterrorism.

Short Bio:
Professor Vincenzo Loia received B.S. degree in computer science from University of Salerno , Italy in 1985 and the M.S. and Ph.D. degrees in computer science from University of Paris VI, France, in 1987 and 1989, respectively. From 1989 he is Faculty member at the University of Salerno where he teaches Safe Systems, Situational Awareness, IT Project & Service Management. His current position is as Chair and Professor of Computer Science at Department of Management and Innovation Systems. He is the coeditor-in-chief of Soft Computing and the editor-in-chief of Ambient Intelligence and Humanized Computing, both from Springer. He is an Associate Editor of various journals, including the IEEE Transactions on System, Man and Cybernetics: Systems; IEEE Transactions on Fuzzy Systems; IEEE Transactions on Industrial Informatics; IEEE Transactions on the IEEE Transactions on Cognitive and Developmental Systems. His research interests include soft computing, agent technology for technologically complex environments Web intelligence, Situational Awareness He was principal investigator in a number of industrial R&D projects and in academic research projects. He is author of over 390 original research papers in international journals, book chapters, and in international conference proceedings. He hold in the last years several role in IEEE Society in particular for Computational Intelligence Society (Chair of Emergent Technologies Technical Committee, IEEE CIS European Representative, Vice-Chair of Intelligent Systems Applications Technical Committee).

Speech Title:    Adversarial Attacks and Defenses for Multi-view Learning

Abstract:
Deep models are highly susceptible to adversarial perturbations. Even if the benign examples are added with imperceptible perturbations, most of the models will output incorrect results. Utilizing the vulnerability of deep models to adversarial examples, adversaries can easily perform malicious attacks, which poses security concerns. Consequently, the robustness of deep models against adversarial attacks has become a crucial area of research. Although various kinds of single-view adversarial attacks and defense have been proposed, there is no specific research on the adversarial attacks and defenses for multi-view deep models. As we all know, multi-view models usually have superior performance. However, it is an open problem whether multi-view models are more robust to adversarial examples than single-view models. We investigate the relative robustness of the multi-view deep model and single-view model, propose effective multi-view adversarial attack and defense methods, and discuss some possible research thoughts on multi-view defenses.

Short Bio:
Shiliang Sun is a Professor with the School of Computer Science and Technology and the Head of the Pattern Recognition and Machine Learning Research Group, East China Normal University, Shanghai, China. His current research interests include kernel methods, multi-view learning, learning theory, approximate inference, and their applications. His research results have expounded in 100+ publications at peer-reviewed journals and conferences. Prof. Sun is on the Editorial Board of multiple international journals, including Pattern Recognition, Information Fusion and IEEE Transactions on Neural Networks and Learning Systems.

Speech Title:    Principles and Practice of Neural Machine Translation

Abstract:
Machine translation is the use of a computer to translate text from one language to another language automatically. The machine translation systems can be divided to ruled-based machine translation, statistical machine translation and neural machine translation (NMT). Recent years, neural machine translation has become the mainstream model due to its superior performance and simple architecture. It has been widely deployed in various commercial translation systems, such as Google Translate, Microsoft Translator, Sogou Translator, DeepL Translator, etc. As an important and challenging research direction in the field of natural language processing (NLP), the techniques involved in NMT are also widely used in other NLP directions such as chatbot, knowledge graph, and question answering systems.
In this tutorial, we will introduce the basic principles of neural machine translation, and then discuss the recent advances and challenges in neural machine translation, including the architecture of NMT, low-resource NMT, constrained decoding for NMT and the application of pretrained language models in NMT. Finally, we will share some research resources for NMT, such as publically available datasets and open source codebases.

Short Bio:
Yun Chen is an assistant professor at Shanghai University of Finance and Economics. She received the Ph.D. degree from The University of Hong Kong in 2018 and the bachelor degree from Tsinghua University in 2013. She was a visiting scholar with the Tsinghua Natural Language Processing Group and NYU CILVR Group. She has published dozens of research papers on top conferences and journals such as ACL，EMNLP，AAAI，IJCAI，ICLR, etc.
Currently, she is the reviewer for ACL, EMNLP, AAAI, COLING, AACL, ACM TALLIP, JCST.