I had the honor to be a Keynote speaker at the 2019 GE Edge and Controls Symposium, September 10-12, 2019. This meeting brought together many top engineers and business leaders from within GE, and also from other companies, national labs, universities, and government. My talk was entitled “Leveraging Machine Learning for Advancing Smart-X Systems and Control“. My key theme was how we can think about the tremendous advances in machine learning and connect them to tools, techniques, design principles, and knowledge-base from systems and control field. It was very well received and there were many thought provoking questions. I also had a chance to reconnect with some friends and collaborators! Slides can be downloaded by clicking on the link above.
News
Oxford-UIDP Summit
I participated in several meetings at the Oxford-UIDP summit held at the Oxford University, UK, July 30 – August 2, 2019. There were two workshops prior to the summit on the themes of the Future of Work and the Future of University-Industry Consortia. I gave a presentation on the Future of Work and Jobs. There is tremendous interest in this topic throughout the world. With advances in automation, machine learning and artificial intelligence, it is becoming increasingly clear that some jobs will disappear, some new job categories will be created, and a large swath of jobs will be transformed. Some tasks will be taken over by machines and algorithms allowing human beings to focus on higher level tasks that require dealing with human-human interactions, uncertainty, and creativity.
Summit focused much of the discussion on issues of connecting university research to societal needs to benefit people. This is an increasingly important concern. Presentations by Sir Mark Walport and Dr. Walt Copan both emphasized this topic from their government roles. UIDP is at the center of this issue in as much as such societal benefits can be realized by connecting universities with for-profit as well as non-profit private sector organizations.
Convergence Research Paradigm
I was invited to give a Key Note talk at the SUNY Research Council meeting on August 6, 2019 in New York city. It was a wonderful opportunity to discuss the potential of this trans-disciplinary research paradigm to address complex problems that lie beyond any single discipline. The meeting was attended by senior leaders from the SUNY system and its many campuses. My host was Grace Wang, Senior Vice Chancellor for Research and Economic Development of the SUNY System and Interim President of the SUNY Polytechnic. There were numerous questions on the key challenges that need to be overcome to realize the potential of convergence paradigm. I have posted my presentation on this website.
I also participates in a symposium entitled “Fostering the Culture of Convergence” at the National Academies of Sciences, Engineering and Medicine earlier this year. I also reviewed the report from this workshop from the National Academies Press which has now been published. It can be downloaded from the NAP website.
Presentations at the Workshop on Learning and Control
There was a Workshop on Learning and Control at IIT Mandi, India in July 2019. It was organized by Professor M. Vidysasagar, IIT Hyderbad. I was invited to give four lectures at this workshop. There were about 35 students and total of about 50 participants. I gave expository lectures on deep learning, generative adversarial networks, and reinforcement learning from control and decision systems perspective. I think the confluence of machine learning and control is likely to be a fertile area for the coming years. We are at the very early stages of developing visions, frameworks, and technical tools, techniques and foundations. As such, my views and opinions are evolving. The discussions and questions were thought provoking. Lecture slides are now under the Presentations tab of this website.
Machine Learning and Control: Memory Augmented Neural Networks for Control
Deepan Muthirayan and I have been working on the role of external memory to improve the performance of adaptive controllers in our work at the confluence of machine learning and control. The basic motivation is that memory plays a very big role in learning. However, by and large, when it comes to control, memory is taken to be the state of the (nonlinear) dynamic controller. Our hypothesis is that by explicitly adding external memory to the adaptive control of uncertain, time-varying dynamical systems, the adaptive performance of the closed-loop system can be improved. We have written two papers where we explore this theme. Our results show the promise and potential of this idea. These can be found on the Publications page of this website.
- Memory Augmented Neural Network Adaptive Controllers: Performance and Stability. [Abstract: “In this paper, we propose a novel control architecture, inspired from neuroscience, for adaptive control of continuous-time systems. The proposed architecture, in the setting of standard neural network (NN) based adaptive control, augments an external working memory to the NN. The external working memory, through a write operation, stores certain recently observed feature vectors from the hidden layer of the NN. It retrieves relevant vectors from the working memory to modify the final control signal generated by the controller. The use of external working memory is aimed at improving the context thereby inducing the learning system to search in a particular direction. …”]
- Memory Augmented Neural Network Adaptive Controller for Strict Feedback Nonlinear Systems. [Abstract: … “We propose a novel backstepping memory augmented NN (MANN) adaptive control method for the control of strict feedback non-linear systems. Here, each NN, in the backstepping NN adaptive controller, is augmented with an external working memory. The NN can write relevant information to its working memory and later retrieve them to modify its output, thus providing it with the capability to leverage past learned information effectively and improve its speed of learning. We propose a specific design for this external memory interface and show that the proposed control design achieves bounded stability for the closed loop system. We also provide substantial numerical evidence showing that the proposed memory augmentation improves the speed of learning by a significant margin.”]
EmTech Next by MIT Technology Review
I am just returning from Boston where Meera Sampath and I participated in the EmTech Next Conference organized by the MIT Technology Review. The theme of the conference was automation and work. We gave a presentation on our work on Socially Responsible Automation (which came out earlier in the NAE Bridge Magazine a few months ago.) Lots of excellent presentations at this conference touching on various aspects of automation, machine learning, artificial intelligence, jobs, tasks, work, education, reskilling, regional and community impacts, role of regulation and policy, etc. It is indeed a very big topic, likely to become even more important. Our slides are posted at the Presentations tab.
New presentations added
I had the honor to present a talk on the future of smart cities for societal benefits at the National Academy of Engineering Regional Meeting at the Illinois Institute of Technology, Chicago on April 24, 2019. I have added my slide deck to the Presentations section of this website.
New draft paper: Socially Responsible Automation: A Framework for Shaping the Future
Dr. Meera Sampath, my colleague and friend, and I have written a paper to address the how we might deal with the impact of automation on jobs and work. We define and articulate Socially Responsible Automation, a conceptual framework that we believe offers an excellent approach to these issues.
Abstract: Technologists have the power and the responsibility to ensure that the tremendous potential of automation and artificial intelligence are harnessed to not just drive economic benefits but to also promote human and societal wellbeing. We present the vision, concept and framework of Socially Responsible Automation to guide the evolution of current and future approaches to automation as practiced in industry. We ground this framework on established principles and methodologies from a number of domains including business ethics, socio-technical systems design and innovation strategy.
Advancing Systems and Control Research in the Era of ML and AI
Our paper Advancing Systems and Control Research in the Era of ML and AI, coauthored with M. A. Dahleh, MIT, has been accepted for publication in Annual Reviews in Control.
Advancing Control in the Era of ML and AI
I have been working on a brief essay entitled “Advancing Control in the Era of ML and AI”. The goal is to lay out the opportunity for the control systems community in the coming years as machine learning and artificial intelligence research progresses. This is a working paper and so comments and suggestions are welcome.