Biswajit Paria

(Bengali: বিশ্বজিৎ পড়িয়া)

Email:
Resume | Google Scholar | Twitter

I currently work as a software engineer at Google, Mountain View in the Ads team. I graduated from Carnegie Mellon University with a PhD (2017-2022) in Machine Learning, advised by Barnabás Póczos and Jeff Schneider. My research interests span active optimization and design of experiments, which includes Bayesian optimization, bandits, and some aspects of sequential decision making. I am also interested in some applications of deep learning including time series forecasting, and neural bandits. During my PhD, I was fortunate to have spent two summers as research interns at Snap Research (2018), and Google Research (2020).

Prior to CMU, I graduated from the Indian Institute of Technology Kharagpur with a 5-year bachelors and masters (2012-2017) in Computer Science and Engineering. I was advised by Pabitra Mitra for my undergraduate thesis.

Outside of work, I am a regular boulderer (≤ V3), and sometimes paint on my iPad.


Papers

(*denotes alphabetical ordering of authors)

  1. Be Greedy – a Simple Algorithm for Blackbox Optimization using Neural Networks
    *Biswajit Paria, Barnabas Poczos, Pradeep Ravikumar, Jeff Schneider, Arun Suggala
    RealML @ ICML, 2022 [paper, work in progress]

  2. A Top-Down Approach to Hierarchically Coherent Probabilistic Forecasting
    *Abhimanyu Das, Weihao Kong, Biswajit Paria, Rajat Sen
    pre-print, 2021 [arxiv, under review]

  3. An Experimental Design Perspective on Model-Based Reinforcement Learning
    Viraj Mehta, Biswajit Paria, Jeff Schneider, Stefano Ermon, Willie Neiswanger
    International Conference on Learning Representations, 2022 [arxiv, paper]
    Preliminary version at EcoRL @ NeurIPS, 2021

  4. Hierarchically Regularized Deep Forecasting
    Biswajit Paria, Rajat Sen, Amr Ahmed, Abhimanyu Das
    pre-print, 2021 [arxiv, under review]

  5. Cost-Aware Bayesian Optimization via Information Directed Sampling
    Biswajit Paria, Willie Neiswanger, Ramina Ghods, Jeff Schneider, Barnabás Póczos
    Real World Experiment Design and Active Learning @ ICML, 2020 [paper]

  6. Autonomous Discovery of Battery Electrolytes with Robotic Experimentation and Machine Learning
    Adarsh Dave, Jared Mitchell, Kirthevasan Kandasamy, Han Wang, Sven Burke, Biswajit Paria, Barnabás Póczos, Jay Whitacre, Venkatasubramanian Viswanathan
    Cell Reports Physical Science, 2020 [paper]

  7. Minimizing FLOPs to Learn Efficient Sparse Representations
    Biswajit Paria, Chih-Kuan Yeh, Ian E.H. Yen, Ning Xu, Pradeep Ravikumar, Barnabás Póczos
    International Conference on Learning Representations (ICLR), 2020 [paper, arxiv, code]

  8. Tuning Hyperparameters without Grad Students: Scalable and Robust Bayesian Optimisation with Dragonfly
    Kirthevasan Kandasamy, Karun Raju Vysyaraju, Willie Neiswanger, Biswajit Paria, Christopher R. Collins, Jeff Schneider, Barnabas Poczos, Eric P. Xing
    Journal of Machine Learning Research (JMLR), 2020 [paper, arxiv, github]

  9. A Flexible Framework for Multi-Objective Bayesian Optimization using Random Scalarizations
    Biswajit Paria, Kirthevasan Kandasamy, Barnabás Póczos
    Uncertainty in Artificial Intelligence (UAI), 2019 (oral) [paper, arxiv]

  10. Analytic Connectivity in General Hypergraphs
    Ashwin Guha, Muni Sreenivas Pydi, Biswajit Paria, Ambedkar Dukkipati
    pre-print, 2017 [arxiv]

  11. Forward Stagewise Additive Model for Collaborative Multiview Boosting
    Avisek Lahiri, Biswajit Paria, Prabir Kumar Biswas
    IEEE Transactions on Neural Networks and Learning Systems (TNNLS), 2016 [paper, arxiv]

  12. A Neural Architecture Mimicking Humans End-to-End for Natural Language Inference
    Biswajit Paria, K. M. Annervaz, Ambedkar Dukkipati, Ankush Chatterjee, Sanjay Podder
    pre-print, 2016 [arxiv]