Research Area : Artificial Intelligence, Machine Learning, Bayesian Learning, Deep Learning
Specific Interests : Bayesian deep learning, Continual learning, Bayesian non-parametrics, Stochastic processes, Gaussian processes, Neural Ordinary differential Equations, Inference algorithms, uncertainty quantification, spatio-temporal and generative modelling.
Applications : Computer vision, natural language processing, social network analysis, astrophysics, autonomous navigation
Teaching machines learn the human way !
My research interest lies in developing machine learning and artificial intelligence algorithms inspired by the way human learning works. Towards this end, we use Bayesian learning, deep learning, continual learning, neural networks, stochastic processes, and differential equations to develop novel machine learning and deep learning algorithms. We develop probabilistic machine learning and deep learning models and algorithms for problems from varied domains of artificial intelligence such as computer vision and natural language processing and application areas such as social networks, web, astrophysics, autonomous navigation, smart mobility etc.
Parametric and non-parametric Bayesian models allow the incorporation of prior information and domain knowledge. Non-parametric Bayesian models such as Gaussian processes additionally allow one to learn rich and flexible models due to their non-parametric nature and allow the model complexity to be determined by the data. This helps to overcome the problem of model selection to a great extent. On the other hand, deep learning models help in effective representation learning and good generalisation performance. I work on developing learning algorithms which combines the best from both the worlds for e.g. Bayesian deep learning and deep Gaussian processes. Another line of research is developing continually evolving models based on point process and neural ordinary differential equations. I am also interested in analysing time series and textual data arising from various application domains. Current research applies Bayesian reasoning and probabilistic modelling to diverse problem domains such as computer vision, language processing, social networks, traffic data, astrophysical data etc. We intend to develop learning algorithms and models which are useful not only for artificial intelligence and computer science but also in general for problems arising in other areas of science and engineering.
Kindly find more details of our work and group at Bayesian Reasoning And INference (BRAIN)
*** Co-organizing the Asian Conference on Machine Learning (ACML) 2022 in Hyderabad, India. More details here
*** Grateful to Accenture for unrestricted research grant.
*** Co-organizing Natural Language Processing session in the AIML vertical in Vaibhav Summit 2020. More details here
*** DST funding for two projects, Towards Developing Next-generation Deep Learning, and Machine Learning for Astrophysical Data Analysis. Ph.D., M.Tech (RA) and Project Assistant positions available ! Students interested in long term internship may also contact.
*** Project with Nvidia on Bayesian deep learning for computer vision and autonomous driving.
*** Accenture funding on gaining business insights by learning from alternative data sources.
*** Organised IITH and RIKEN-AIP joint workshop on Artificial Intelligence during March 15-16, 2019 at IIT Hyderabad.
*** Talk on convolutional deep Gaussian processes at IIT Bombay and AIRC, Tokyo
*** Talk on deep Gaussian processes at the 3rd Indian workshop on machine learning during july 1-3, 2018
*** JICA funding for the Bayesian Deep Learning project in collaboration with Dr. Emtiyaz Khan from RIKEN Centre for Advanced Intelligence project, Tokyo.
***Talk at the Workshop on Analysis and Inference from Unstructured Data during Oct 9-10 at Indian Institute of Science on Social Network Analysis using Point Processes.
***SERB funding for social network analysis project using point processes, deep learning and topic modelling.