PR. AMIR HUSSAIN
Trustworthy Artificial Intelligence-enabled Transformative Healthcare Technologies: Practical Case Studies, Challenges and Opportunities
Amir Hussain obtained his BEng (1st Class Honours) and PhD from the University of Strathclyde in Glasgow in 1992 and 1997 respectively. He is founding Director of the Centre of AI and Data Science at Edinburgh Napier University, UK. His research interests are cross- disciplinary and industry-led, aimed at developing cognitive data science and trustworthy AI technologies to engineer smart industrial and healthcare systems of tomorrow. He has (co)authored three international patents and over 500 publications, including 200+ journal papers and 20 Books/monographs. He has led major national and international projects, and supervised over 35 PhD students. He is the founding Chief Editor of Springer’s Cognitive Computation journal and Springer Book Series on Socio-Affective Computing. He has served as invited Associate Editor/Editorial Board member for various other journals, including the IEEE Transactions on Neural Networks and Learning Systems, Information Fusion (Elsevier), the IEEE Transactions on Systems, Man and Cybernetics: Systems, and the IEEE Transactions on Emerging Topics in Computational Intelligence. Amongst other distinguished roles, he is a member of the UK Computing Research Committee (national expert panel of the IET and the BCS), General Chair of IEEE WCCI 2020 (the world’s largest IEEE CIS technical event in computational intelligence, comprising IJCNN, IEEE CEC and FUZZ- IEEE), and Chair of the IEEE UK and Ireland Chapter of the IEEE Industry Applications Society.
SENIOR MEMBER, IEEE
Airborne wind energy
Received the B.S. degree in electrical engineering from the Lebanese University Beirut Lebanon in 2004 and the M.S. degree and the Ph.D. degree in automatic control from the Grenoble Institute of Technology Grenoble France in 2005 and 2007 respectively.
After one year as a Researcher with the LIRMM Laboratory Montpellier France, he joined the University Grenoble Alpes Grenoble as an Associate Professor with the Grenoble Image Parole Signal Automatique Laboratory (GIPSA-Lab) Automatic Control Department.
Now he is the head of the COPERNIC research team in the data science pole of Gipsa-lab and head of the Master MARS of UGA.
His research interests include electrical vehicle integration nonlinear control predictive control for energy systems and airborne wind-energy systems.
In France, the energy transition law imposes a reduction of the fossil energies down to 50% within 10 years. Transition scenarios foresees renewable part of energy production to reach between 40% and 70%, with an increasing portion of wind energy. The conventional wind energy sector will have diﬃculty meeting this strong demand for renewable energy that stems from the need to drastically reduce the ecological impact of our society. In ﬂoating oﬀshore wind turbines in particular, the amount of materials per unit of power involved in the construction of wind turbines and ﬂoating devices may render this concept unsustainable if it is scaled commensurably with current fossils energy production levels. The emerging airborne wind energy sector offers breakthrough concepts that will allow diversifying the wind energy production oﬀer. Today, in the ﬁeld of high altitude wind energy also known as Airborne wind energy (AWE, in French Éolienne aéroportée), a thriving multi-disciplinary community of researchers and technologists in academia and industry all over the world is well-established . The delivered research results are gradually assessing and eliminating feasibility risks and improving the understanding of AWE systems, ultimately bringing these concepts closer and closer to industrialization. The claimed advantages are lower installation and maintenance costs, higher capacity factors, higher ﬂexibility and better adaptation to ﬂoating platforms with respect to the current established renewable technologies. AWE industry is moving from small scale hand-controlled prototype drones and kites to full scale autonomous controlled commercial systems. For example, in August 2019 Makani operated a 600kW energy kite prototype from a ﬂoating platform in the North Sea. AWE systems can be divided into two classes depending on the location of energy production: On-board production systems (as Makani) and on-ground production systems that have two phases: production and recovery (or consumption) phases. In this talk, after reviewing the main aspects of airborne wind energy, some recent research activities will be described, including the design and construction of a small-scale prototype, the development of dynamical models and algorithms for the wing. Finally, the design and some successful experimental testing of an automatic control algorithm will be presented. At the end of the talk, future research directions and needs will be also discussed.
NIDHAL CARLA BOUAYNAYA
ASSOCIATE DEAN AND PROFESSOR OF ELECTRICAL AND COMPUTER ENGINEERING ROWAN UNIVERSITY
Towards Robust and Self-Aware Machine Learning - A Bayesian Framework for Uncertainty Propagation in Deep Neural Networks
- Professor of Electrical and Computer Engineering with expertise in statistical learning and data science.
- Director of Rowan Artificial Intelligence Lab (RAIL).
- Leads the academic research and graduate studies mission and portfolio as the Associate Dean for Research and Graduate Studies.
Deep neural networks (DNNs) have surpassed human-level accuracy in various fields, holding the promise of emerging technologies, such as self-driving cars and autonomous unmanned aircraft systems, smart cities infrastructure, personalized treatment in medicine, and cybersecurity. However, unlike Humans who have a natural cognitive intuition for probabilities, DNN systems – being inherently deterministic – are unable to evaluate their confidence in the decisions. To truly deserve its name, an artificial intelligence system must be aware of its limitations and have the capacity for insightful introspection.
This talk will advance Bayesian deep learning methods that are able to quantify their uncertainty in the decision and self-assess their performance, are robust to adversarial attacks, and can even expose an attack from ambient noise. This talk will establish the theoretical and algorithmic foundations of uncertainty or belief propagation through complex deep learning models by adopting powerful frameworks from optimal estimation problems in non-linear and non-Gaussian dynamical systems.
The challenge in DNNs is the multi-layer stages of non-linearities in deep learning models, which makes propagation of high-dimensional distributions mathematically intractable. Drawing upon powerful statistical frameworks for density propagation in non-linear and non-Gaussian dynamical systems, we introduce Tensor Normal distributions as priors over the network parameters and derive a first-order Taylor series mean-covariance propagation framework. We subsequently extend this first-order approximation to an unscented framework that propagates sigma points through the model layers. The unscented framework is shown to be accurate to at least the second-order approximation of the posterior distribution. We finally learn the entire predictive distribution using Particle Filtering, a powerful class of numerical methods for the solution of optimal estimation problems in non-linear, non-Gaussian systems.
The uncertainty in the output decision is given by the propagated covariance of the predictive distribution. We show that the proposed framework performs an automatic logit squeezing, which leads to significantly enhanced robustness against noise and adversarial attacks. Experimental results on benchmark datasets, including MNIST, CIFAR-10, real-world synthetic aperture radar (SAR), and Brain tumor segmentation (BraTS 2015), demonstrate: 1) superior robustness against random noise and adversarial attacks; 2) self-assessment through predictive confidence that monotonically decreases with increasing levels of ambient noise or attack; and 3) an ability to detect a targeted attack from ambient noise.