Machine Learning India bio photo

Machine Learning India

Fostering data science and machine learning in India

Email Mailing List Twitter Github

Are you a student, professor, CEO or Maschinenmensch? Subscribe to ml-india's google group to join the discussion and recieve updates, news and resources about India's ml-ecosystem. Click here

Nitin is an assistant professor at IIT Kanpur. He graduated from IIT Kanpur in 2004 with a B.Tech. in computer science and an interest in biology. He then joined the Ph.D program in bioinformatics and systems biology at the University of California, San Diego (UCSD). After completing his PhD in 2009, he spent a few months in a cognitive neuroscience lab at UCSD, and then moved to the National Institutes of Health as a post-doc to work on electrophysiology. He specializes in systems neuroscience, olfaction and computational biology. His research interest lies in understanding the fundamental mechanisms of neural circuits for processing information.


Nitin on..


ML-India: How did you get into the field of neuroscience?

Nitin: I had always been interested in the philosophical aspects of how we think and how we are able to remember things. Initially, it was just a curiosity. But while I was pursuing my B.Tech in computer science, I got exposed to biology through some coursework and projects. I enjoyed these and also started reading about the Human Genome project. I wanted to learn more, so I opted for graduate studies in Bioinformatics- a subject which is deeply connected to computer science. During this period, I continued exploring topics in neuroscience . After my Ph.D, I thought I might as well switch fields since I was spending so much time on neuroscience. I got an opportunity to do my post-doctoral studies in neuroscience and I took it. I’ve been into this field ever since.

ML-India: Could you let us know of some of the fundamental problems in neuroscience that are being solved today?

Nitin: Over the years, neuroscience has become a vast field and people are working at many levels to understand the brain. One could categorize the studies into three levels. One is at the ‘molecular/cellular’ level. People conducting studies at this level are trying to understand the composition of our nervous system and answer questions like “What are neurons made of?” and “What are the individual chemical molecules present in neurons that determine how they function?”. Another is at the ‘circuits and systems’ level, where people are trying to understand the circuits connecting neurons. Consider the functioning of the visual system when someone sees an object - light travels to the retina, from which the information then goes through a bunch of nerves to the primary visual cortex V1. It then moves to V2, V3 and so on. People have been interested in understanding this underlying complex structure and how information is communicated in such systems. The third level is ‘cognitive neuroscience’ where people are trying to broadly model the functioning of higher level activities like decision making and thinking. Researchers are investigating if there exists a correlation between specific functions of the brain’s parts and the behavior displayed.

This subject can also be categorized as fundamental and applied. Researching chemicals that trigger certain responses in the brain, or chemicals that can be used to manipulate the brain’s activity to relieve someone from a disorder, is an example of applied neuroscience. However, most of the research that is happening is in the fundamental domain.

ML-India: Have we had any major breakthroughs in the last 20-30 years?

Nitin: In terms of basic understanding, yes, we’ve gotten many new insights. The biggest breakthrough came around 50 years ago which gave a clearer understanding of the activity of neurons. Hodgkin and Huxley, who got the Nobel prize for this work, answered how information is converted into electrical pulses and determined the equations of such a transfer while considering the concentration of ions inside and outside the neuron. The visual system was the first to be studied in detail. Hubel and Weisel, again Nobel laureates, analyzed how the visual cells perceived an image, not as a whole, but as a set or combination of dots and lines. In recent years, a lot of work has been done using FMRI. Over the past 20 years or so, people have identified broader areas in the brain that are responsible for different activities. Another area of work has been in identifying the connections between different neurons in the brain, and how those connections exist. There are millions of neurons in an animal’s brain, and billions in a human one. A lot of work towards building the experimental tools to study them is being done. And this has been a major challenge for neuroscientists. Our ability to detect neuron activity and analyze its existence using non-invasive methods has improved vastly but is still very limited. One of the more recent developments has been in optical recording and optogenetics. Traditionally, people use electrodes, metal wires or glass capillaries with some electrolyte to read neuron activity, but this is invasive in nature and can only read a small number of neurons. More recently, scientists have started using optical imaging, where a chemical is put into the brain which changes its fluorescence whenever an activity takes place in that region. One can then use just a microscope to image the activity of a large number of neurons. Academics have also developed proteins that can change the activity of specific neurons by illuminating them.

ML-India: In this canvas, how would you place your work?

Nitin: I’m currently working on the ‘molecular’ and ‘systems’ levels. The majority of work in my lab is being done on olfactory systems. That can be described as work done in systems neuroscience. We’re trying to analyze how different smells are processed in the brain. We are using insects as our model systems because they are easy to experiment on, they have a very well developed sense of smell and their olfactory systems are similar in organization to that in humans. Specifically, I’m trying to understand how certain smells become attractive or repulsive to mosquitoes. It has been observed that mosquitoes use our body’s smell as one of the cues for finding us. In addition to sensing heat levels and using their visual features, smell plays an important role. Different smells of people can determine whether mosquitoes will bite them. Our work might help in producing more effective repellents which are less harmful to humans.

ML-India: How does your training in computer science come into the picture?

Nitin: There are many ways in which computer science is useful. One area is the analysis of experiments. Experimental tools generate a lot of data and one needs the knowledge of data analysis techniques to make sense of it. For instance, neuroscience routinely involves electrical recording, which generates data for neurons over a period of time. Techniques like feature detection, noise removal from the data, deconvolution, prediction using machine learning, unsupervised clustering, visualization and other handy tools from computer science and electrical engineering immediately follow. For instance, while recording electrical activity in neurons, we can sometimes put the electrode outside the neuron to pick up activity from other neighboring neurons as well. Within a single trace, we obtain a mixed activity of neurons. Analyzing such traces would typically involve the usage of some of the techniques mentioned above.

The other area is modeling or simulations of neuro-systems. Based on the knowledge that we have about how neurons function, we can make a computer representation of a neural circuit. Each neuron can be represented by a node whose voltage can be governed by a certain set of differential equations. The neurons are connected based on the considered configuration and we can then provide it input and observe the overall activity of the network to look for any patterns that may be occurring. This kind of modeling predicts the behavior of such specific networks.

ML-India: What are some of the other projects going on in your lab?

Nitin: We recently started working in the applied cognitive area. In this work, we are not doing fundamental discoveries but are using what is known to alleviate a disease. For instance, people who have depression use medication for relief, and some people use psychotherapy. Although therapy has no side effects, there is a lot of stigma around it which makes people reluctant to go to a therapist. As a result, researchers have tried to develop a computer-based method for delivering therapy to patients. This would reduce the problems of inaccessibility of a therapist while overcoming the stigma. We are working on such systems with some renowned therapists so that the therapy can be delivered efficiently. This will involve some traditional content of therapy for the patients along with some interactive features to make it more personalized. This will eventually also involve machine learning to match and personalize interventions with the patients’ personalities. Once developed, this software will be made available on our website- treadwill.org.

ML-India: It seems from this conversation that experimental analysis trumps theoretical studies in neuroscience. Which is unlike the case in, say, computer science.

Nitin: Yes, that is accurate. Neuroscience is at a stage where we are still severely limited by experimental data. I think that our understanding of different parameters of neurons and of different types of neurons is very limited. We have a long way to go before we have sufficient data on various parts of the brain.

ML-India: You talked about the human genome project. Any other grand initiatives which are happening in the field of neuroscience?

Nitin: There are a few big initiatives happening. One is called the BRAIN initiative by the US government which is trying to accelerate brain research through innovative neurotechnology. The idea, again, is to focus on developing new techniques using recent tools of nanotechnology and in developing optical techniques to record and manipulate the activity of the neuron. Another is a Europe-wide project called the Human Brain Project. It is partly dedicated to developing new technologies and is partly into developing computer models to incorporate all the experimental data that exists currently. There is a group at Janelia research farms, which is a campus of the Howard Hughes Medical Institute, that has set up an institute to focus primarily on neuroscience. They are using the Drosophila (a small fly) model system to map most of the neurons in the brain. They are using genetic tools to examine neurons’ structure and activities and their role in determining various behaviors in animals. I think this will become very powerful in the coming years.

ML-India: Your thoughts on the current state of the art being pursued in India? Are there any fundamental problems being chased down and hunted by the groups in India?

Nitin: I think neuroscience is a growing field. It is really small in India in comparison to the traditional fields of biochemistry and biophysics. There are a few academics doing very good work. For example, Prof. Upinder Bhalla at NCBS is working on memory. At my institute, IIT Kanpur, there are two other people working in this field. Dr. S Ganesh is working on the molecular and cellular aspect of the Lafora disease, a form of epilepsy, and he is trying to understand and identify the causal proteins. Dr. Jonaki Sen, who is into developmental neuroscience, is studying how neurons develop and form connections as an animal grows in age.

ML-India: Where do you think India can improve? Any low hanging initiatives that you think can we can target to push the envelope in neuroscience?

Nitin: Since the neuroscience community in India is very small currently and geographically dispersed, it is certainly desirable to hold regular meetings and discussions to come up with interesting new ideas. In terms of projects, I think science is very global today, so I don’t think there are any low hanging initiatives specifically for India. In applications, there are a few areas where adaptation to the Indian context will be very useful. Our project to deliver computerized behavioral therapy for depression is an example. Funding is not a problem in India currently. One problem that I feel is the lack of high-quality students. I have been fortunate to get really good students, but in general, the number of applicants we get is small. Many good students take up industrial jobs or go abroad. Also, my kind of work requires students who are comfortable with engineering concepts as well as have an interest in biology, and it is not easy to find such students.

ML-India: What are the future steps in your research?

Nitin: We’ll continue working primarily in the areas of olfaction and mosquito behavior. We are thinking of starting some work on genetic modifications of mosquitoes so that we can use optical manipulation tools on them. In the cognitive domain, we may expand to other disorders beyond depression and will try to make our tools smarter.

ML-India: What would you recommend to a non-biologist aiming to enter this field and get a sense of it?

Nitin: There are plenty of online courses available today for students who are interested in neuroscience. One particular course that I liked is ‘Fundamentals of Neuroscience’ taught by a professor at Harvard. Also, labs in India are open to interested people with an engineering background who want to learn more about this field and are willing to spend some time. Year-long projects at these labs will very helpful to students who are looking to get a cross-disciplinary experience.

ML-India: Any closing thoughts?

Nitin: Engineering students and professionals who are interested in biological fields should not hesitate in exploring this as a career option. People usually think that it will be really difficult to succeed in this field since they haven’t studied biology in high school and college, and that switching fields won’t be a wise decision to take. I knew nothing about biology before I switched fields. I learned everything on the fly, and speaking from personal experience, I think that it’s easier to study biology after engineering than going the other way. So people need not be afraid of entering this field. Also, it is a really good time to be in neuroscience research. With some great research happening in this field, students and researchers will learn a lot!

ML-India: Thank you for your time, Nitin. Best wishes from ML-India for your future work.