Kalyanmoy Deb is University Distinguished Professor at Department of Electrical and Computer Engineering in Michigan State University, USA. Prof. Deb's research interests are in evolutionary optimization and their application in multi-criterion optimization, modeling, and machine learning. He was awarded IEEE Evolutionary Computation Pioneer Award for his sustained work in EMO, Infosys Prize, TWAS Prize in Engineering Sciences, CajAstur Mamdani Prize, Distinguished Alumni Award from IIT Kharagpur, Edgeworth-Pareto award, Bhatnagar Prize in Engineering Sciences, and Bessel Research award from Germany. He is fellow of ACM, IEEE, and ASME. He has published over 640 research papers with Google Scholar citation of over 228,000 with h-index 145. He is in the editorial board on 10 major international journals. More information about his research contribution can be found from https://www.coin-lab.org.
Title: Machine Learning Assisted Multi-Criterion Optimization
Abstract: With the advancement of machine learning methods, various predictive and feature learning tasks are being solved in a better and more confident manner than before. In this talk, we use recent machine learning methods to improve the performance of evolutionary multi-criterion optimization algorithms for finding a well-distributed near-Pareto optimal solution. The ideas will be supported with experimental results on test and certain engineering design problems.
Professor, Department of Electrical Engineering and Information Systems
The University of Tokyo, Japan
Akira Hirose received the Ph.D. degree in electronic engineering from the University of Tokyo in 1991. In 1987, he joined Research Center for Advanced Science and Technology (RCAST), the University of Tokyo, as Research Associate. In 1991, he was appointed as Instructor at RCAST. From 1993 to 1995, on leave of absence from the University of Tokyo, he joined the Institute for Neuroinformatics, University of Bonn, Bonn, Germany. He is currently Professor with the Department of Electrical Engineering and Information Systems, the University of Tokyo. The main fields of his research interests are wireless electronics and neural networks. In the fields, he published several books such as Complex-Valued Neural Networks, 2nd Edition (Springer 2012). He served as the Founding President of Asia-Pacific Neural Network Society (APNNS) (2016), President of Japanese Neural Network Society (JNNS) (2013-2015), Vice President of the IEICE Electronics Society (ES) (2013-2015), Editor-in-Chief of the IEICE Transactions on Electronics (2011-2012), Associate Editor of journals such as the IEEE TRANSACTIONS ON NEURAL NETWORKS (2009-2011), IEEE GEOSCIENCE AND REMOTE SENSING NEWSLETTER (2009-2012), Chair of the Neurocomputing Technical Group in the IEICE, Founding Chair of the IEEE CIS NNTC Complex-Valued Neural Network Task Force (2010-), Governing Board Member of APNNA/APNNS (2006-), IEEE Geoscience and Remote Sensing Society (GRSS) All Japan Chapter Chair (2013-2015) and IEEE Computational Intelligence Society (CIS) All Japan Chapter Chair (2017- 2018). He also served as the General Chair of Asia-Pacific Conference on Synthetic Aperture Radar (APSAR) 2013 Tsukuba, International Conference on Neural Information Processing (ICONIP) 2016 Kyoto, and International Geoscience and Remote Sensing Symposium (IGARSS) 2019 Yokohama. Dr. Hirose is a Fellow of the IEEE and the IEICE, and a member of JNNS and APNNS.
Title : Quaternion reservoir computing for spatiotemporal PolSAR earth observation
Abstract :
Quaternions hold the ability to represent polarization information as a unified entity. They are widely used in polarimetric synthetic aperture radar (PolSAR) data processing. However, the functions are limited to statistical treatments so far. With the increasing amount of PolSAR data, it is becoming more important to estimate temporal variations as well as spatial characteristics. In this keynote, we discuss quaternion reservoir computing (QRC), which we recently proposed, in particular in spatiotemporal PolSAR data analysis. Experiments deal with time-series PolSAR data obtained by Advanced Land Observation Satellite 2 (ALOS-2) of Japan Aerospace Exploration Agency (JAXA) for agricultural sites and tropical rainforests to detect anomaly. The accuracy achieved with QRC is 97%, whereas it is 90% with conventional recurrent neural networks (RNNs) and 94% with real-valued reservoir computing (RC). The results suggest that our proposed QRC is effective in various PolSAR applications such as land use and disaster management in the spatiotemporal domain.
Professor of Computer Science and Oden Institute of Computational Engineering and Sciences
The University of Texas at Austin, USA
Prof. Bajaj’s current research is on the computer science and computational mathematics foundations of statistical and dynamic decision making, learning dynamical systems and exploratory visualization. He develops machine learning and geometric optimization algorithms for applications in the physical, chemical, health and environmental data sciences, medicine and neuro-morphic computing. Prof. Bajaj revisits his past research in modeling multi-scale form and function, through the lens of inferential data sciences. Prof. Bajaj’s current research projects include (a) forward and inverse optimization problems in microscopy, spectroscopy, electro-magnetic and electro-optical wide spectrum imaging; (b) generative shape and new material design for spatially realistic and phenomenological models; (c) learning from nature and generating models and mechanisms with accelerated and emergent properties. He also teaches an undergraduate course titled “Geometric Foundations of Data Sciences” and a graduate course titled “Predictive Machine Learning”. Prof. Bajaj has courtesy appointments and supervise undergraduates, M.S. and Ph.D. students from several UT departments, including biomedical and electrical engineering, neuroscience, and mathematics.
The following is a link to his most cited publications:
http://scholar.google.com/citations?user=gyL3CZ0AAAAJ"&user=gyL3CZ0AAAAJ
Title: Progressive AI for multi-modality diagnosis and individualized intervention in the treatment of Parkinson's disease
Abstract: Parkinson's disease (PD) is a progressive neurodegenerative disorder with varied symptoms and diverse progression rates amongst humans. Today, the accuracy and efficiency of PD diagnosis is considerably enhanced through a modern suite of wearable sensing devices and equipped with progressive and real-time AI models. In this talk I shall investigate biological and imaging differences between PD patients who test positive or negative, in the Cerebral Spinal Fuild Seed Amplification Assay (CSFSAA) test. A transformer-based segmentation ML model is trained to extract volumetric data and dopamine levels across 108 regions of interest (ROI) from T1-MRI and DaT-SPECT (DATSCAN) imaging respectively. In addition using a newly developed streaming compositional Bayesian co-clustering that jointly models tissue-specific diffusion tensor imaging (DTI) biomarkers and clinical assessments, permits effective uncovering of multimodal patterns that are predictive of disease severity. I shall also describe how wearable devices with Inertial Measurement Units (IMU's) combined with time-series analysis of arm swing and human gait provides real-time monitoring of symptoms, allowing for dynamic adjustments to treatment plans. Unlike traditional methods treating time series as static vectors or fixed sequences one uses Motion Code that views each time series, regardless of length, as a realization of a continuous-time stochastic process.
Distinguished Engineer, NVIDIA Corporation, Singapore
Jeff is a HPC specialist with over 30 years of experience in developing and optimising scientific codes and architecting HPC solutions. Jeff’s primary area of expertise is in Weather & Ocean modelling, having previously worked at the New Zealand Oceanographic Institute (now NIWA), Toyota Motor Corporation, and on FEA/CFD analysis for America’s cup class yachts for Team New Zealand. Prior to joining NVIDIA, Jeff spent 16 years working at SGI on designing and deploying several operational Weather centres across the Asia Pacific region. Jeff leads the Earth Systems Science research team for the NVIDIA AI Technology Centre’s joint laboratory established at Nanyang Technological University (NTU) in Singapore. There he works on collaborations with academia and industry partners in the fields of accelerated Climate/Weather research and in the application of AI to Earth Systems Modelling. Jeff holds a Postgraduate Diploma in Computer Science from the University of Auckland, and is currently a PhD candidate at the University of Newcastle on Tyne.
Title: Developments in Generative AI for Climate Science
Abstract: Climate change is the defining problem of our time and an existential threat to humanity. Recent increases in climate change-attributed extreme weather events underscores the need for better forecasts. A key goal of the climate and weather community is to provide the best possible predictions to policy makers, but long-term prediction is inherently challenging to get right, and uncertainties can limit action to fight climate change. Current estimates are that the required traditional computational capabilities will not be feasible before 2060 at the earliest – and we can’t wait that long! Artificial Intelligence gives us a new and powerful method to substantially accelerate the processing time for climate projections. Recent advances in machine learning, particularly in the field of Generative AI, have provided new tools to advance our capabilities in climate science. In this talk, we will discuss various applications for AI/ML in Climate and weather modelling and present the latest breakthrough research in GenAI for Climate Science.
Partner Group Engineering Manager, Microsoft India Development Center, India
Samir Jain works as a Partner Group Engineering Manager in Microsoft India Development Center. He leads Azure networking IDC Hyperscale networking connectivity area comprising of overall Microsoft Wide Area Network, Optical and Edge. He also leads Azure Sovereign Cloud Networking area. His team is designing, building and running world’s largest network comprising of routers, optical devices, fibres, ISP circuits and the software services to manage the network. This network spans 140 countries, 60+ Azure regions, 200+ Azure data centres, 190+ Azure edge sites connecting to 4000+ ISPs and 160,000 miles of Optical fibres. He has 28+ years of experience primarily in networking area across L1-L7 protocols, embedded systems to cloud, broadband network to hyper-scalers. He is deeply passionate about how technology can help to improve the lives of millions of human beings. He co-founded a social enterprise, BodhaGuru, to help government school children get an affordable quality education using technology. He has filed 6 patents, published 2 IEEE papers and co-authored new SSL VPN protocol (SSTP).
Title: Designing Cloud scale network to meet AI needs and using AI to run it
Abstract: In this talk, Samir will share how they are designing the massive network to meet the growing needs of scale, security and performance to support AI workloads. He will also share the technical challenges to design & run this network and how they are leveraging AI to solve some of the challenges.