Publications

Journals :

• Abstract Operations Research Modeling Using Natural Language Inputs

with Junxuan Li, Ryan Wickman, Sahil Bhatnagar, Arko Mukherjee

MDPI, Information 2025.

• vqSGD: Vector Quantized Stochastic Gradient Descent

With Venkata Gandikota, Daniel Kane, Arya Mazumdar

IEEE Transactions on Information Theory (TIT), vol. 68, no. 7, 2022.

• Communication-Efficient and Byzantine-Robust Distributed Learning  

     With Avishek Ghosh, Swanand Kadhe, Arya Mazumdar, Kannan Ramchandran

       IEEE Journal on Selected Areas of Information Theory (JSAIT), Sep. 2021.

Conference:

• FED-CURE: A Robust Federated Learning Algorithm with Cubic Regularized Newton                Link 

with Avishek Ghosh, Arya Mazumdar.  

IEEE International Symposium on Information Theory (ISIT), 2024.  

•  vqSGD: Vector Quantized Stochastic Gradient Descent .      Link   

With Venkata Gandikota, Daniel Kane, Arya Mazumdar . 

       International Conference on Artificial Intelligence and Statistics (AISTAT),2021.  

• Distributed Newton Can Communicate Less and Resist Byzantine Workers.     Link 

 With  Avishek Ghosh , Arya Mazumdar.          

          Neural Information Processing Systems (NeurIPS), 2020

• A Workload-Adaptive Mechanism for Linear Queries Under Local Differential Privacy.                                   Link   

        With Ryan McKenna, Arya Mazumdar, Gerome Miklau.

       Proceedings of the VLDB Endowment (VLDB), 2020.

• High Dimensional Discrete Integration over the Hypergrid.                   Link   

       With  Arya Mazumdar,  Soumyabrata Pal . 

        Uncertainty in Artificial Intelligence (UAI), 2020.

• Communication Efficient Distributed Approximate Newton Method. 

        With Avishek Ghosh, Arya Mazumdar, Kannan Ramchandran.  

        IEEE International Symposium on Information Theory (ISIT), 2020.

• Communication-Efficient and Byzantine-Robust Distributed Learning           Link   

With Avishek Ghosh, Swanand Kadhe, Arya Mazumdar, Kannan Ramchandran. 

IEEE International Symposium on Information Theory (ISIT), 2020.

• Robust Gradient Descent via Moment Encoding with LDPC Codes.               Link   

 With  Ankit Singh Rawat, Arya Mazumdar.  

        IEEE International Symposium on Information Theory (ISIT), 2019 .

• Shaping Proto-Value Functions Using Rewards. 

       With  Chandrashekar Lakshminarayanan, Sindhu Padakandla, Shalabh Bhatnagar.  

        European Conference Artificial Intelligence 2016

Workshops and Talks:

• FED-CURE: A Robust Federated Learning Algorithm with Cubic Regularized Newton

with Avishek Ghosh, Arya Mazumdar.

Workshop of Federated Learning and Analytics in Practice, co-located with 40th International Conference on Machine Learning 

• Estimation of Shortest Path Covariance Matrices Link                

With  Cameron Musco.

Presented at WOLA workshop 2020.

• vqSGD: Vector Quantized Stochastic Gradient Descent .

With Venkata Gandikota, Daniel Kane, Arya Mazumdar . 

      Appeared as short paper ICML 2019 Workshop on Security and Privacy in Machine Learning.  (Oral Presentation)

•  A Practical Method for Workload-Aware Frequency Estimation Under Local  Differential Privacy

       With Ryan McKenna, Arya Mazumdar, Gerome Miklau

       Workshop on Theory     and Practice of  Differential      Privacy (TPDP), 2019.)

• Robust Gradient Descent via Moment Encoding with LDPC Codes.

       With  Ankit Singh Rawat, Arya Mazumdar. 

       Extended abstract in The SysML Conference, 2018. Oral Presentation.