报告题目：Real-time Classification of Big Data Stream through Hashing Techniques
Lianhua Chi is a lecturer in the Department of Computer Science and Information Technology at La Trobe University, Australia. She completed her dual Ph.D. in Machine Learning and Data Mining at University of Technology Sydney (UTS), Australia, and Huazhong University of Science and Technology (HUST), China in 2015. She joined La Trobe after almost three years at IBM Research Australia where she was working as a Postdoc on Watson Education and Health. Before she joined IBM Research Australia, she was a Data Specialist at University of New South Wales (UNSW) and a Visiting Researcher at University of Technology Sydney (UTS). Her main research interest is in making sense of big data with effective, agile and interactive data analytics, especially in Health Care area. Currently, she is working on AI on Health. She has received several awards due to her research contribution, such as "Best Paper Award", "Top 200 Young Researchers Globally", “Young Global Changer”, "IBM External Honors" and “Romberg Grant Award”. She has published more than 20 Research papers in International Conferences and Journals and also filed 3 patents in these or related areas. In 2018, she was invited to visit Harvard Medical School for potential research collaborations.
Many organisations, such as financial firms, are today actively trying to extract meaning from an explosion of structured and unstructured data: social media streams, smartphone data, videos. The fundamental challenge of big data is the enormous volume, dynamic change, and need for real-time response. When handling big data, our existing frameworks severely lack scalability and real-time capacity, mainly because big data are treated as a static gigantic set, whereas data from applications such as social networks and surveillance videos are typically updated every single minute.
For decades, hashing is one of the most effective tools commonly used to compress data for fast access and analysis. Hashing techniques have also evolved from simple randomisation approaches to advanced adaptive methods considering locality, structure, or label information of the data for effective hashing. This talk will review and categorise existing hashing techniques and further introduce how hashing techniques enhance big data stream classification (including structured data and unstructured data) based on my research work on this area.
邀请人： 杜博 教授