KEYNOTE SPEAKER
Zhouchen Lin, Peking University.
Speech Title:
Discovering Symmetry in Data
Abstract:
Inductive bias is key to the success of a learning system. Symmetry is an important kind of inductive bias and has been widely applied in problems such as the design of equivariant networks and the discovery of governing equations. However, in complex scenarios symmetry is not known in advance. In this talk, I will present techniques that use Lie group and Lie algebra to discover symmetry in data.
Bio of the Keynote Speaker Zhouchen Lin is a Boya Special Professor of Peking University and a vice-dean of School of Intelligence Science and Technology. His research interests include machine learning and optimization. He has published over 400 technical papers and 5 monographs, receiving over 47,000 Google Scholar citations. He has been Area Chairs and Senior Area Chairs of CVPR, ICCV, ICML, NIPS/NeurIPS, AAAI, IJCAI and ICLR for many times. He was a Program co-Chair of ICPR 2022 and an associate editor of IEEE Transactions on Pattern Analysis and Machine Intelligence (IEEE TPAMI). He is currently a member of ICML Board of Directors, an Associate Editor in Chief of the IEEE TPAMI, and an associate editor of the International Journal of Computer Vision, and Acta Automatica Sinica. He is a Fellow of the CCF, the CSIG, the AAIA, the IAPR and the IEEE. He is also a recipient of The Chinese National Science Fund for Distinguished Young Scholars and a principal investigator of a State Key Project for AI.
Bio of the Keynote Speaker Zhouchen Lin is a Boya Special Professor of Peking University and a vice-dean of School of Intelligence Science and Technology. His research interests include machine learning and optimization. He has published over 400 technical papers and 5 monographs, receiving over 47,000 Google Scholar citations. He has been Area Chairs and Senior Area Chairs of CVPR, ICCV, ICML, NIPS/NeurIPS, AAAI, IJCAI and ICLR for many times. He was a Program co-Chair of ICPR 2022 and an associate editor of IEEE Transactions on Pattern Analysis and Machine Intelligence (IEEE TPAMI). He is currently a member of ICML Board of Directors, an Associate Editor in Chief of the IEEE TPAMI, and an associate editor of the International Journal of Computer Vision, and Acta Automatica Sinica. He is a Fellow of the CCF, the CSIG, the AAIA, the IAPR and the IEEE. He is also a recipient of The Chinese National Science Fund for Distinguished Young Scholars and a principal investigator of a State Key Project for AI.
Zhiwen Yu, Harbin Engineering University
Speech Title:
Crowd Sensing 2.0: From Human-Centered to Heterogeneous Crowd Sensing
Abstract:
Crowd sensing is a new sensing paradigm that uses individual mobile and ubiquitous sensing capability to accomplish complex sensing tasks. In this speech, I will introduce our work in crowd sensing in various aspects, such as theory, methods, and platform. Furthermore, I will present the main idea of crowd sensing 2.0, including the features and enabling technologies of heterogeneous crowd sensing.
Bio of the Keynote Speaker Dr. Zhiwen Yu is currently a vice president of Harbin Engineering University, China and a professor of Northwestern Polytechnical University, China. He has worked as an Alexander Von Humboldt Fellow at Mannheim University, Germany from Nov. 2009 to Oct. 2010, a research fellow at Kyoto University, Japan from Feb. 2007 to Jan. 2009, and a post-doctoral researcher at Nagoya University, Japan in 2006-2007. His research interests cover Ubiquitous Computing, Internet of Things, and Crowd Sensing and Computing. He is the Editor-in-Chief of CCF Transactions on Pervasive Computing and Interaction. He has served as an associate/guest editor for a number of international journals, such as IEEE Transactions on Human-Machine Systems, IEEE Communications Magazine, and ACM Transactions on Intelligent Systems and Technology. He received the CCF Young Scientist Award in 2011, the Humboldt Fellowship in 2008, and the CCF Excellent Doctoral Dissertation Award in 2006. He got the National Science Fund for Distinguished Young Scholars in 2017.
Bio of the Keynote Speaker Dr. Zhiwen Yu is currently a vice president of Harbin Engineering University, China and a professor of Northwestern Polytechnical University, China. He has worked as an Alexander Von Humboldt Fellow at Mannheim University, Germany from Nov. 2009 to Oct. 2010, a research fellow at Kyoto University, Japan from Feb. 2007 to Jan. 2009, and a post-doctoral researcher at Nagoya University, Japan in 2006-2007. His research interests cover Ubiquitous Computing, Internet of Things, and Crowd Sensing and Computing. He is the Editor-in-Chief of CCF Transactions on Pervasive Computing and Interaction. He has served as an associate/guest editor for a number of international journals, such as IEEE Transactions on Human-Machine Systems, IEEE Communications Magazine, and ACM Transactions on Intelligent Systems and Technology. He received the CCF Young Scientist Award in 2011, the Humboldt Fellowship in 2008, and the CCF Excellent Doctoral Dissertation Award in 2006. He got the National Science Fund for Distinguished Young Scholars in 2017.
Dave Towey, University of Nottingham Ningbo China
Speech Title:
Thinking Differently to Shape our Future Digital Society: Metamorphic Testing, Thinking and Teaching
Abstract:
Recognising correctness is not always easy. When testing software, knowing
whether or not its output or behaviour is correct would appear essential to
effective testing: Yet this is not always possible.Advances in computing technology
-including generative AI and big data -have revealed deficiencies in how we apply
quality assurance (QA) to modern software. Software testing is an important QA
process: We provide the software with test inputs, and its behaviour or output
is checked for correctness. An inability to identify the correctness is called
the oracle problem. Software facing the oracle problem has been called untestable,
because traditional software testing methods may not be usable.
Metamorphic testing (MT) provides a new way to deal with untestable software. Rather than focusing on individual output correctness, it looks at relationships and relations that should hold if the software has been properly implemented: A violation of such relations indicates a fault in the software.
In addition to introducing the principles of MT, and some of its significant successes, this interactive session will also explore the application of MT ideas more generally to the teaching and learning of software testing, and beyond.
Bio of the Keynote Speaker Dave Towey received his BA and MA degrees from The University of Dublin, Trinity College; PgCHE from University of Nottingham; PgCertTESOL from The Open University of Hong Kong; MEd from The University of Bristol; and PhD from The University of Hong Kong. He was the first foreign academic recipient of the Zhuhai municipal outstanding teacher award, in 2007, and he received the Lord Dearing award in 2017 for his outstanding contribution to the development of teaching and student learning. He is a full professor, and the head of the School of Computer Science, which he joined in September 2013. He also serves as deputy director of the International Doctoral Innovation Centre (IDIC). He was previously the associate dean for education and student experience for the Faculty of Science and Engineering.After graduating from The University of Dublin, Trinity College, Dave worked in Japan in the late 1990s, helping develop a breast cancer screening tool using ultrasound imaging technology and fuzzy reasoning. After this, he lived in Hong Kong from 2000 to 2005/2006, where, as well as completing his PhD, he worked at The University of Hong Kong, and as a teacher and teacher trainer in the local school system. In 2005, he became involved in a newly created liberal arts college in Zhuhai, Mainland China, the Beijing Normal University - Hong Kong Baptist University United International College (UIC), where he remained until 2013. While at UIC, he taught modules related to computer science, linguistics, and education. He also held several roles, including deputy director of the English Language Centre, and coordinator (director) of the Teaching English as a Second Language (TESL) degree programme. In these roles, he oversaw delivery of a large number of pre-service and in-service training courses and workshops.Dave's research interests span a number of areas, including technology-enhanced teaching and learning, and software testing, especially metamorphic testing and adaptive random testing.
Metamorphic testing (MT) provides a new way to deal with untestable software. Rather than focusing on individual output correctness, it looks at relationships and relations that should hold if the software has been properly implemented: A violation of such relations indicates a fault in the software.
In addition to introducing the principles of MT, and some of its significant successes, this interactive session will also explore the application of MT ideas more generally to the teaching and learning of software testing, and beyond.
Bio of the Keynote Speaker Dave Towey received his BA and MA degrees from The University of Dublin, Trinity College; PgCHE from University of Nottingham; PgCertTESOL from The Open University of Hong Kong; MEd from The University of Bristol; and PhD from The University of Hong Kong. He was the first foreign academic recipient of the Zhuhai municipal outstanding teacher award, in 2007, and he received the Lord Dearing award in 2017 for his outstanding contribution to the development of teaching and student learning. He is a full professor, and the head of the School of Computer Science, which he joined in September 2013. He also serves as deputy director of the International Doctoral Innovation Centre (IDIC). He was previously the associate dean for education and student experience for the Faculty of Science and Engineering.After graduating from The University of Dublin, Trinity College, Dave worked in Japan in the late 1990s, helping develop a breast cancer screening tool using ultrasound imaging technology and fuzzy reasoning. After this, he lived in Hong Kong from 2000 to 2005/2006, where, as well as completing his PhD, he worked at The University of Hong Kong, and as a teacher and teacher trainer in the local school system. In 2005, he became involved in a newly created liberal arts college in Zhuhai, Mainland China, the Beijing Normal University - Hong Kong Baptist University United International College (UIC), where he remained until 2013. While at UIC, he taught modules related to computer science, linguistics, and education. He also held several roles, including deputy director of the English Language Centre, and coordinator (director) of the Teaching English as a Second Language (TESL) degree programme. In these roles, he oversaw delivery of a large number of pre-service and in-service training courses and workshops.Dave's research interests span a number of areas, including technology-enhanced teaching and learning, and software testing, especially metamorphic testing and adaptive random testing.
Sizhen Bian, NWPU
Speech Title:
Biomimetic Intelligence: Neuromorphic Sensing for Embodied AI
Abstract:
Artificial intelligence is advancing faster than academic research cycles. Training compute for frontier models has grown by roughly ten orders of magnitude since 2010, general-purpose foundation models are absorbing entire fields such as natural language processing and computer vision, and the frontier of AI has shifted decisively from academia to industry. Yet scale has not bought everything: today's AI remains far from the human brain's roughly 20-watt efficiency, struggles with embodied perception and action, and learns from far more data than any animal. These are exactly the areas where biological intelligence excels, suggesting a distinctive role for academia: not to out-scale industry, but to decode the mechanisms of biological intelligence.
This talk argues that biomimetic intelligence should learn mechanisms rather than outputs, and that intelligence does not begin in the brain but in sensing. Drawing on our work on the Kraken event/frame processing SoC, the ColibriUAV neuromorphic nano-drone platform and low-power event-based eye tracking, I will show that the first bottleneck of closed-loop neuromorphic systems is often not the spiking network but the information stream that feeds it. Biological sensors such as the retina, cochlea, skin and vestibular system each use task-specific codes, and our evaluation of rate, time-to-first-spike, binary and delta encodings on inertial sensor signals shows that the choice of encoding determines accuracy, robustness and energy efficiency. I will conclude with five design principles for neuromorphic sensing in embodied AI and outline adaptive, task-driven encoding as the next research question: building artificial nervous systems that start at the sensory edge.
Bio of the Keynote Speaker Sizhen Bian is a Professor in the School of Computer Science at Northwestern Polytechnical University (NWPU), China. He received his Ph.D. in Embedded Intelligence from TU Kaiserslautern, Germany. He was a Postdoctoral Researcher at ETH Zurich where he worked on neuromorphic edge platforms such as the Kraken SoC. His research spans neuromorphic sensing and computing, ubiquitous computing, wearable and textile sensing, sensor-based human activity recognition, edge AI and human-computer interaction. He has authored or co-authored more than 70 peer-reviewed publications and has received five international recognitions including the PerCom'25 Best Workshop Paper Award, IMWUT'23 Distinguished Paper Award, and ECCV 2024 Best Demo Award.
This talk argues that biomimetic intelligence should learn mechanisms rather than outputs, and that intelligence does not begin in the brain but in sensing. Drawing on our work on the Kraken event/frame processing SoC, the ColibriUAV neuromorphic nano-drone platform and low-power event-based eye tracking, I will show that the first bottleneck of closed-loop neuromorphic systems is often not the spiking network but the information stream that feeds it. Biological sensors such as the retina, cochlea, skin and vestibular system each use task-specific codes, and our evaluation of rate, time-to-first-spike, binary and delta encodings on inertial sensor signals shows that the choice of encoding determines accuracy, robustness and energy efficiency. I will conclude with five design principles for neuromorphic sensing in embodied AI and outline adaptive, task-driven encoding as the next research question: building artificial nervous systems that start at the sensory edge.
Bio of the Keynote Speaker Sizhen Bian is a Professor in the School of Computer Science at Northwestern Polytechnical University (NWPU), China. He received his Ph.D. in Embedded Intelligence from TU Kaiserslautern, Germany. He was a Postdoctoral Researcher at ETH Zurich where he worked on neuromorphic edge platforms such as the Kraken SoC. His research spans neuromorphic sensing and computing, ubiquitous computing, wearable and textile sensing, sensor-based human activity recognition, edge AI and human-computer interaction. He has authored or co-authored more than 70 peer-reviewed publications and has received five international recognitions including the PerCom'25 Best Workshop Paper Award, IMWUT'23 Distinguished Paper Award, and ECCV 2024 Best Demo Award.