Varun Aggarwal, co-founder and CTO of Aspiring Minds and a major proponent of machine learning in India, offered excellent views on how India can turbocharge AI research. Aggarwal, who has authored Leading Science and Technology: India Next?, and is the person behind ml-india.org, wrote a whitepaper which highlighted that India needs PhD students studying AI, as well as, world-class research institutes. India should strike beyond derivative research and focus on India-based solutions, emphasised Aggarwal in his whitepaper.
Even though NITI Aayog has undertaken a slew of initiatives and is establishing programmes to bolster research and development in India, the country’s academia has traditionally lagged behind in ML and research. In an earlier post — Where does India stand as machines become intelligent? — Aggarwal had stated that Indian academia produced fewer ML papers than a single university in China. In fact, India’s contribution to ML in the last 15 years, as evident from this graph from ML India below, is dismal as compared to countries like China, US, UK and Germany. India’s total research output is 745 as compared to China’s 3,956. When it comes to industrial research, the scenario is even grimmer, notes Aggarwal.
Aggarwal is a recognised expert on developing and applying statistical learning and optimisation techniques to real-world problems, including circuit modelling, design, parallel processing and network coding. A prolific patent-filer — he has five US patents pending and his work has been published in more than 20 technical publications. His work for the algorithms he developed for Aspiring Minds is also celebrated. He is the founder of the MIT India Reading Group, a platform for collaborative research on socioeconomic issues in India. At Aspiring Minds, he has developed world’s first ML-based programming assessment, the world’s first automated test of motor skills and the only scalable way of spoken English assessment.
This article lists key insights from Aggarwal’s paper. They clearly state that AI is a wholesome opportunity for India, but can only be realised through research and innovation that can be converted into sustainable business models.
Research In ML Should Be The Building Block For AI Strategy Over The Next 5 Years
India Needs A Critical Mass Of AI Researchers: According to Aggarwal, India should attract 500 top AI researchers to have a critical mass — a move that requires several government interventions, such as creating fellowships for 500 AI professors for a period of 10 years. To attract the best talent, Aggarwal proposes that the fellowships should not be just restricted to Indian nationals but also be opened to foreign nationals to draw the best minds to the country.
Generating 2,000 PhD Students Over The Next 5 Years: Researchers can’t succeed alone, Aggarwal emphasises. To that end and also to rekindle India-centric research, he proposes that the country should strive to create a pool of 2,000 top class PhD students to drive the Indian AI ecosystem forward.
Interdisciplinary Centres Can Strengthen Industry-Researcher Relationship: From hosting seminars to industry interactions and collaborations, interdisciplinary centres can help in advancing industry-academia partnership. It will also narrow down on India-specific problems, set up infrastructure and create a holistic programme for the development and maintenance of AI tools. The centres can also focus on developing differentiated datasets for specific problems.
Universities Should Create A Unique Research Agenda: From setting up an internal committee to focusing on goals and benchmarks to crafting unique AI programmes, Aggarwal, a prolific researcher says that universities should avoid the “me too” approach and develop unique research agendas. The universities should try to raise the bar when it comes to disruptive research and also include industry stakeholders who can guide the research program.
Not Just Universities, School Bodies Like NCERT And CBSE Should Take A Lead In Including AI In Curricula: We also spoke about this perspective, in view of China’s growing heft in AI and how the country has mandated AI in its school curricula. In this case, Aggarwal alludes to his own initiative for kids — Datasciencekids.org, a fun platform for introducing computer science to children.
AIM had also noted that to counter the talent gap, India should start training the next generation in subjects that can help build a foundation for AI — subjects like Maths with a special emphasis on statistics, probability, algebra and calculus. Universities should include these subjects — programming, designing, robotics, engineering, computer architecture, database management, cryptography and program analysis, among others — at both the undergraduate and graduate level.
Try deep learning using MATLAB