Featured Eminent Speaker of ACM India: Amod Jai Ganesh Anandkumar
The ACM India Eminent Speaker Program (ESP) provides ACM Student and Professional chapters and institutional partners in India with direct access to top technology leaders, innovators and researchers who will give talks on contemporary and engaging issues that are important to the computing community. Chapters can invite ESP speakers to give talks as part of events that they hold. ACM India will help cover the costs for travel while the local chapter will arrange accommodation and local logistics. While most of the talks should be in-person to encourage interactions, some of the talks may be done virtually. For more details, see the ACM India ESP Page
Bio
Dr. Amod Anandkumar is a Director of Research at Harman International, a Samsung group company. He leads a multinational research team working on efficient deep learning for advanced mobility and audio systems. Prior to this, he was the Director of Curriculum and Applications at Skill-Lync (a 2023 Top YCombinator ed-tech startup). Amod has previously held leadership and research roles at MathWorks, Samsung, and GE, working on AI, autonomous vehicles, wireless networks, and biomedical engineering applications.
Amod holds a PhD from Loughborough University, UK, and a BTech from NIT Karnataka. Amod has delivered keynotes, invited talks, and tutorials at numerous national and international conferences, seminars, and tradeshows. He has also served on Board of Studies and advisory committees at academic institutions. Amod is a Senior Member of IEEE and an Executive Committee member of IEEE Signal Processing Society Bangalore Section.
Please tell us a little about the research group you lead at Harman International.
HARMAN International is a subsidiary of Samsung and develops connected automotive and consumer electronics products under brands such as JBL, Harman Kardon, AKG, etc. HARMAN X is the research division of Harman International working on long-term innovation and technology development. I’m proud to lead a diverse multinational team in HARMAN X whose expertise spans deep leaning, computer vision, signal processing, software engineering and systems. We work on Edge AI, developing optimized deep learning-based solutions that are deployed on a variety of hardware and software platforms. This involves interdisciplinary research across areas such as deep learning theory, neural network architecture design, model compression, statistics, and software optimization.
How did your career path lead you to research on Edge AI?
When I was an undergraduate student, I was fascinated by signal processing and worked on a research project on blind source separation for speech and music signals. This motivated me to pursue a PhD at the intersection of applied signal processing, wireless communication and optimization techniques such as game theory. But by the end of my PhD, I saw that my research contributed to the theory of wireless systems but was far away from practical implementation. I wanted my research to be closer to practical solutions and this motivated me to transition to industry research.
Over the course of my stints at GE, Samsung, and MathWorks, I worked on developing solutions across healthcare, consumer electronics, automotive and aerospace applications. The common thread across this decade of industry experience was that algorithms don’t exist in isolation but are parts of large complex systems that have to work under constraints of the underlying hardware and software platforms. This led me to my current role at Harman, where my applied research team focuses on finding and solving problems in the translation from fundamental research to practical technology.
What are some applications of your work in Edge AI?
HARMAN’s automotive electronics solutions consists of the full hardware-software-application stack for solutions in infotainment, digital cockpit, in-cabin sensing, car audio, connectivity, and telematics. The consumer electronics division has many audio products such as headphones, earbuds, wireless speakers, and soundbars. Most of the new AI solutions that we develop go into one or more of these products. The AI models operate on audio, bio-signal, and/or video data in real time and are deployed on mobile phones, embedded devices, and automotive SoCs. On the audio side, a typical application is to understand the audio content for improving the rendering of soundtracks. On the automotive side, applications include smart customizations and interior monitoring to improve the in-cabin experience and safety.
What is an emerging area of research in this field that will have a significant impact in the coming years?
I’m very excited about our current work on estimating human mental and physiological states using various bio-signals. In this project, we are combining deep insights from applied psychology and psychophysics with the power of deep learning on large scale datasets to design mental state detectors and vital sensing models. For example, we’re able to learn patterns in micro-fluctuations of skin tone from video data to predict the heart rate of an individual. Similarly, we’re able to track eye gaze and blink patterns to detect whether the individual is alert, distracted, stressed or sleepy. We’re initially targeting the automotive domain for developing driver distraction, drowsiness, and vital signs monitoring solutions.
Looking at the trajectory of progress in AI, sensors, and computing platforms, I believe that we will see the immense impact of such interdisciplinary research on human state estimation in the future. I expect that our future cars, smartphones, wearables, and hearables will be able to learn a lot about us and understand our mental, physical, and physiological state. This will help integrate our digital and physical lives more deeply, ultimately leading to improved health and wellbeing.
What motivates you to be a part of the ESP program?
I am passionate about promoting high quality research in India, and I firmly believe that good research has to happen across the country and not just in a few centres of excellence in premier institutions. I think that ACM has an excellent reach all over India through its student and professional chapters. Specifically, the Eminent Speaker Program provides a great and accessible platform to share my experience in R&D with a diverse audience. Also, I have benefited from attending various research talks in my student and early career days, and I would love to use this opportunity to give back and hopefully inspire budding researchers all over India.
If your ACM Chapter would like to invite Amod Anandkumar to give a talk, please follow the guidelines for inviting an ESP Speaker available here.
You can also start an ACM Chapter or become an ACM Member.
