Niloy Ganguly

Professor, IIT Kharagpur

E-mail id: niloy[at]cse.iitkgp.ac.in

 

 

Brief Bio-sketch:

Dr. Niloy Ganguly is a Professor in the Dept. of Computer Science and Engineering at IIT Kharagpur. He is a Fellow of Indian National Academy of Engineering. He spent 2 years as a Research Scientist in Technical University, Dresden, before joining IIT Kharagpur in 2005, and has risen to the rank of Professor in 2014. He has done his Btech from IIT Kharagpur and his Phd from IIEST, Shibpur. His research interests lie primarily in Social Computing, Machine Learning, and Network Science. He has published in 60 journals and 140 conferences, several of which are in reputed international venues such as NIPS, KDD, ICDM, IJCAI, WWW, CSCW, EMNLP, CHI, ICWSM, INFOCOM, Physical Review E, Euro Physics Letter, IEEE and ACM Transactions etc. He has served in the program committee of COMSNETS, NetSciCom, WWW, DEBS and CODS. Prof Ganguly's work has been recognized through awards by NSF, Cisco, NetApp, Samsung, and Yahoo!, among others. He has received prestigious research grants and projects, notably from Data Transparency L

ab, IMPRINT, ITRA, Intel, HPE, Adobe, Microsoft Research, Accenture, BEL, and TCS. He has guided 16 Ph.D. and 5 M.S. students during this tenure. He is the founding member of the Complex Networks Research Group (CNeRG), comprising faculty members, research scholars, and other students affiliated to the department. The group is a success story in itself, with several long-standing impactful collaborations, and presence in reputed venues across domains such as Social Computing, Machine Learning and Deep Learning, Natural Language Processing, Network Science, Networked Systems, etc.

 

NEVAE: A DEEP GENERATIVE MODEL FOR MOLECULAR GRAPHS 

Deep generative models have been praised for their ability to learn smooth latent representation of images, text, and audio, which can then be used to generate new, plausible data. However, current generative models are unable to work with molecular graphs due to their unique characteristics their underlying structure is not Euclidean or grid-like, they remain isomorphic under permutation of the nodes labels, and they come with a different number of nodes and edges. In this paper, we propose NeVAE, a novel variational autoencoder for molecular graphs, whose encoder and decoder are specially designed to account for the above properties by means of several technical innovations. In addition, by using masking, the decoder is able to guarantee a set of valid properties in the generated molecules. Experiments reveal that our model can discover plausible, diverse and novel molecules more effectively than several state of the art methods. Moreover, by utilizing Bayesian optimization over the continuous latent representation of molecules our model finds, we can also find molecules that maximize certain desirable properties more effectively than alternatives.