Ramanarayan Mohanty works as a research scientist at PCL India-Intel Labs, where he develops graph-based machine learning and deep learning algorithms.

His research interests primarily include Graph AI, machine learning, deep learning and computer vision.

INDIAai interviewed Ramanarayan to get his perspective on AI.

How did your interest in algorithms begin?

While pursuing my master's at IIT Kharagpur, I came across some of my seniors pursuing their PhDs and working on machine learning algorithms. During our tea time, we had some discussions, which motivated me to work on algorithms for my PhD.

Could you tell us about your professional and academic background?

I completed my graduation in computer science and engineering from the Biju Pattnaik University of Technology. Then I pursued my master's (MS Research) in embedded systems and PhD in Graph machine learning from the Indian Institute of Technology Kharagpur. After completing my PhD, I joined one company called Pathpartner Technology as a Lead Research Engineer. Then I switched to Intel Labs, where I work as a Research Scientist.

What differences and similarities do you notice between being a university researcher and a research scientist?

In India, the significant differences between an academic researcher and a research scientist in the industry are as follows:

i) Industrial research scientist has a very well-defined problem from day one to work on, but for a university researcher, the problem statement is not that clear and well-defined; with time, it gets clear.

ii) As research scientists, we have an abundant number of resources in terms of machines, clusters and whatever we need, we get it immediately, but as university researchers, we don't have that luxury.

iii) Another significant difference is that in the industry, most of the research outcomes are directly applied in real-life scenarios or real applications. Still, in university research, this is not always the case. 

Similarly, we majorly go for journals and conferences in industry and academia. Both in industry and university, we do research that is both theoretical as well as application-oriented.

Tell us about the problems you solved during your PhD research and your chosen topic.

During my PhD research, I worked on classification problems of hyperspectral images in adverse and challenging environments. One problem I worked on is the "Class Identification and discrimination issue" in hyperspectral image classification. Hyperspectral images are captured by satellites or UAVs containing hundreds of spectral bands. Each image includes some square kilometres of space on the earth, including houses, roads, forests, water bodies, green fields, etc. 

Each entity is called one class, each with a unique spectral signature. The spectral signatures are the major discriminating feature of hyperspectral images. However, sometimes two classes have the same signature, like the road and roof of a building have the same spectral signature as they are made up of cement. That creates confusion in classifying the images, called the "class discrimination" problem. 

Similarly, sometimes one image class has two different spectral signatures due to light variation. This problem is called the "class identification" problem. To solve these issues, I used some graph machine learning-based techniques by considering both the spectral and spatial features. Similarly, I had worked on some other interesting problems in that domain.     

Tell us about your role as a Research Scientist at Intel India Parallel Computing Labs. What is your daily routine?

At Intel, the work culture is very flexible. As a Research Scientist at Intel Labs, my daily routine involves the following:

• Reading some research papers or blogs.
• Coding.
• Running experiments and analyzing results.
• Attending some meetings and doing some project planning.

Also, I spend some time at the foosball table and the cafeteria to relax my mind. Overall, the day is quite enjoyable and, at the same time, productive.   

The global computer vision market was worth USD 11.22 billion in 2021 and is expected to grow at a compound annual growth rate (CAGR) of 7.0% between 2022 and 2030. In addition, artificial intelligence (AI) in computer vision technology is gaining popularity in various applications. What are your thoughts on this market trend?

Computer vision (CV) with deep learning (DL) is one of the most progressive and rapidly growing fields in research and applications. The advancements in CV with DL research are immediately adapting in the commercial world. In current scenarios, CV and DL are massively transforming the sectors like security, retail, robotics, manufacturing, automotive, healthcare, etc. Predictably, this commercial usage of CV is just the tip of the iceberg. The full potential use of the CV is yet to come. The rapid development in augmented and VR applications like the metaverse makes CV more intuitive. It integrates into people's lives by letting them interact with real objects in the virtual world. The future of CV and DL will be significantly impacted by the new architectures, advanced cloud solutions for handling massive data and automated solutions to reduce the research to market time further. The recent advancement in the transformer architecture in CV and cloud computing services will give more boost to scale up the CV and DL solutions on a population scale with enhanced capability. It will provide an upward thrust to this trend. 

What advice do you have for those who want to work in AI research? What are the most efficient methods of progress?

Currently, AI is a scorching topic; everyone wants to pursue a career in AI and data science. Hence, the competition is also very high. So to make a career in AI, students need to look into two basic things: proper programming skills and a fundamental understanding of machine learning concepts. From a programming skill perspective, they need to be proficient in any programming language (e.g. Python mostly prefered). From an ML concept perspective, they must know four pillars (Linear algebra, probability, statistics and Calculus). 

I believe learning AI by practising is the most efficient method of progress. Either by reading research papers, trying to understand and implement them to reproduce the results or implementing small capstone projects is the most effective way of progress.    

Could you provide a list of notable academic books and journals on artificial intelligence?

There are lots of books and journals on AI available in the market. However, I found some books very interesting, 

"Pattern Recognition and Machine Learning" by Bishop are very good for fundamental machine learning, 

"Dive into Deep Learning" by Alex J. Smola is an excellent deep learning open-sourced interactive book freely available on the internet. 

Apart from this, "Deep Learning" by Goodfellow is also an excellent book to follow. From a research perspective, lots of journal and conference proceedings are available. Some of the most advanced and notable conference proceedings include Neurips, ICLR, ICML, AAAI, CVPR, ICCV, ECCV, etc. The published papers are free, and anyone with an internet connection can access them easily.