Dr. Debanga Raj Neog

Research Expertise

Research Areas in Deep Learning:

  • Deep architectures: Vision Transformers, Generative Adversarial Networks

  • Representation Learning

  • Geometric Deep Learning

Applications of Machine Learning and Deep Learning (ML/DL):

  • Computer vision: Localization, classification, segmentation, and tracking (e.g., eye/face tracking, hand gesture classification/localization/tracking, facial expression classification, medical image segmentation)

  • Computational Imaging: high dynamic range imaging (e.g., video capturing industrial welding)

  • Augmented Reality/Virtual Reality (VR/AR): medical (e.g., anatomy mirroring) and industrial applications (e.g., AR-based monitoring and inspections of welding processes)

  • Computer Graphics: realistic modeling and animation of human facial movements.


Dr.  Arghyadip Roy

Research Expertise

Research Areas in Wireless Communication Networks:

Radio Access Technology (RAT) selection/ User Association

Heterogeneous Networks (4G LTE, 5G/6G, WiFi), Cellular-WiFi Offloading

Software-Defined Networking

Network Coding

Research Areas in Machine Learning:

Markov Decision Process (MDP)

Reinforcement Learning (RL)

Multi-armed Bandits

Risk-sensitive MDP/ RL

Structure-aware RL

Stochastic Approximation

Policy Gradient and Actor-critic Algorithms

Application of RL in Wireless Communication Networks:

Fast, robust and low-complexity RL algorithms for

  • RAT selection and User Offloading

  • Mobile Edge Computing

  • Internet-of-things (IoT)


Dr. Rhythm Grover

Research Expertise:

The primary focus of research is on statistical modeling and analysis of periodic and nearly periodic signals. This field of research is applicable to solve problems in diverse fields of science and engineering such as environmental sciences, communication systems, biomedical, acoustics, finance, stock valuation, image processing, and others.

Current research interests:

1) Development of efficient algorithms for parameter estimation of signal processing models.

2) Derivation of statistical properties of classical parameter estimation methods.

3) Study of robust methods of parameter estimation in presence of outliers in the data.


Dr. Ashish Anand

Research Expertise

Natural Language Processing, Machine Learning and its Applications, Computational Biology.

Focus of the Group

NLP: Fine-Grained Information Extraction, Multilingual NLP, Visual Question Answering, BioNLP
Machine Learning and Its Applications: Noise Aware Models, Data Augmentation, Automated Labeling, Predictive Maintenance
Computational Biology: Systems Network Biology, Computational Genomics.

More Details at:  https://www.iitg.ac.in/anand.ashish


Prof. Prabirkumar Saha

Research Expertise

  •   System Identification - Modelling based on Orthogonal Basis Function, Wiener-Hammerstein Strategy, Neuro-Fuzzy Learning Scheme, Wavelet Network.
  •   Control Systems - Controllability and Stability Analysis, H 2 /H ∞ Optimal Control, Model Predictive Control, Internal Model Control.

Application of AI in control systems (Intended research)
Dynamic systems modeling, structural properties, model reduction, identification, stability, feedback, optimality, robustness,adaptation, fault tolerance, and architecture have been among the central concerns on the theoretical side of ML and AI. These issues have been explored in a wide variety of settings: linear, nonlinear, stochastic, hybrid, distributed, supervisory, and others. The future is just as promising as there are a multitude of directions for future theoretical and applications research.

Some initial ideas for interesting research explorations are briefly outlined below:

Traditionally, control systems analysis and design has been based on detailed mathematical modelling of the processes and with fairly well-understood sources of uncertainty. However, a much larger opportunity arises in areas where such detailed, mechanistic mathematical models do not exist, and/or where the amount of uncertainty is significantly greater with unknown sources.

The key goal of neuroscience and cognitive science has been to build machines that can learn and think for themselves. Cybernetics was conceived as “the scientific study of control and communication in the animal and the machine”. Over time, this connection between control and cybernetics did not develop as fully as the mathematical control theory paradigm. New insights at the confluence of neuroscience, cognitive science, reinforcement learning, and AI to conceptualize new architectures for versatile, intelligent and adaptive controllers might be a good research initiative.

While there have been impressive advances in deep learning, many aspects remain only partially understood. For instance, there is considerable gap in  understanding why deep neural networks have small generalization error in many real world applications. It is well-known fact that saddle points in high-dimensional non-convex optimization are a critical barrier in optimization and training of deep neural networks. There is thus an opportunity for systems and control theory community to contribute innovative non-convex optimization solutions to saddle point problem.)


Prof. Siddhartha Pratim Chakrabarty

Research Expertise

The primary areas of research include Mathematical Finance and Mathematical Biology.
Mathematical Finance:
(1) Computational aspects of pricing financial derivatives.
(2) Portfolio theory.
(3) Financial risk management and modelling.
(4) Actuarial mathematics.
(5) Climate risk and its impact on financial markets.
(6) Data analytics in finance.
Mathematical Biology:
(1) Deterministic, stochastic and control theoretic approaches, in modelling of biological phenomenon.
(2) Mathematical epidemiology and its data driven analysis.



Dr. Biplab Bose

Research Expertise

Dynamical Systems in biology; Emergent Phenomena in Biology; Statistical Physics in Biology.

Focus of the lab: Understanding Design Principles in Biology Through Mathematical Models









Current projects:

  1. Understanding self-organization in planar cell polarity using statistical physics.
  2. Understanding the dynamical aspects of network motifs in biological networks.  
  3. Parameter free analysis of network motifs and networks.
  4. Using machine learning tools in analysis of dynamical systems.

Relevant works:
Percolation in a reduced equilibrium model of planar cell polarity. Physical Review E 100 (3), 032408.
Morphological State Transition Dynamics in EGF-Induced Epithelial to Mesenchymal Transition. J Clin Med. 2019 8(7).


Dr. Souptick Chanda

Research Expertise

Statistical modelling of bone morphology to address ethnic and demographic variations

Phenomenological predictions of bone healing

Orthopaedic implant design and optimization by employing NN-GA based evaluation of failure criteria

Image based detection of pathophysiological phenomena (using machine learning/AI)

Virtual reality based surgical simulation using open-source game engines

Phenomenological prediction of post-operative bone healing based on fuzzy interpretations