Angshuman Paul is an assistant professor at the department of computer science and engineering, IIT Jodhpur.

He received his PhD from the Indian Statistical Institute. Medical imaging is one of his research interests.

INDIAai interviewed Angshuman to get his perspective on AI.

How did an electronics and telecommunications engineer decide to do research in computer science? How does this change take place?

I did my Bachelor of Engineering from Jadavpur University. The course curriculum of Electronics & Telecommunication at Jadavpur University had several significant subjects related to computer science. So, we had a chance to learn a few computer science subjects during our undergrad. Then, during my final year project, I worked on a problem that required the application of computer vision techniques. This problem is the starting point of my interest in computer vision. Subsequently, during my master's, I worked on a computer vision problem that I found extremely interesting. As a result, I decided to do my PhD in computer vision and machine learning. 

What were the initial challenges you encountered while conducting your research?

I faced different challenges in the other initial phases. For example, during my master's and PhD, I had to learn the appropriate techniques when I started working on the relevant research problems. This learning took significant effort. Furthermore, there were very few baseline methods during my postdoc research.

You completed a postdoctoral fellowship at the National Institutes of Health. What differences do you notice in research between Indian universities and abroad? In what ways do Indian universities need to get better?

My National Institutes of Health lab had sufficient computing resources for each researcher. This lab certainly helped in our research. Researchers will benefit from the research in Indian universities if a better infrastructure in terms of relevant resources is created.

According to Globe Newswire, the Global Robotic Medical Imaging Systems Market is to grow by $621.24 million between 2022 and 2026, at a CAGR of 14.1%. What steps should Indian universities take to prepare for this growth?

We should promote medical imaging research by improving collaboration between hospitals, technical institutes, and universities. Building a better computing infrastructure is also required. A significant effort should be directed toward creating indigenous datasets. 

Do you believe that India's artificial intelligence research and education benefit global markets? What are your insights?

Yes, definitely. Education-wise, most of the top technical institutes in India are providing SOTA courses in AI. Moreover, the quality of AI research in India is also excellent. As a result, many Indian candidates enrol for PhDs and postdocs in top universities and institutes worldwide. Furthermore, in most AI industries across the globe, a good number of Indians are working. This transition shows the effectiveness of AI education and research in India.

Today, students can study AI using a variety of online courses and resources. So how do we pick a suitable system or resource?

At the beginner level, one should go through the online materials that sufficiently cover the major mathematical concepts of AI and ML. I think an excellent mathematical foundation is essential in this context. Subsequently, a student should learn the various popular AI techniques and models and the intuitions behind those techniques and designs. Then, one should start implementing those techniques for various benchmark problems in this phase. Finally, based on the interest, researchers should learn strategies for some application areas or focused research areas.    

What advice do you give people who want to conduct AI research in India?

I think an AI researcher would benefit from learning the mathematical foundations of different AI techniques. It is also essential to have good programming skills in the relevant areas. While reading the existing research paper should be a regular practice, one should start developing and implementing independent ideas. Good knowledge about the relevant datasets is also essential.

Could you provide me with a list of essential books and articles on AI research?

1. Online course: Click here
2. Book: Click here
3. Book: Deep learning