Sayar Karmakar
Assistant Professor
Department of Statistics
University of Florida
email: sayarkarmakar at ufl dot edu

Research Interest :
Time series, High-Dimensional Statistics,
Forecasting, Change-point,
VAR models, Econometrics
Gradient Descent methods in Neural Nets
Combinatorial Probability, Networks
Robust Statistics, Posterior consistency

Curriculum Vitae

Google scholar

Upcoming invited talks

Some of my recent and upcoming invited talks are here

Recent Updates

Nov 2022: Our Paper got a revision request from Neurocomputing.

Nov 2022: We submitted a paper on Adversarial attacks on Binarized Neural Network.

August 2022: We submitted a Paper on climate risks and forecastability of the Trading Volume of Gold.

August 2022: Since last January we have started working on game-theory problems of different flavors. We just submitted our first Paper from this collaboration.

July 2022: We submitted Paper where we explore Neuro-tron algorithm and its performance.

July 2022: We submitted Paper which can be viewed a comprehensive analysis after we published our initial analyses on the impact of COVID-19 on cyberbullying in 2020. This involves some new NLP techniques for pre-filtering. We are thankful to Twitter for successfully approving our academic project application and share authentic data.

June 2022: Our Paper has received a revision request from Financial Innvoation.

May 2022: We submitted Paper that explores some theoretical properties of Deep Operator Net. This is my first work in empirical process after longing to work in this area for a while.

April 2022: Our Paper is now accepted at Neural Networks.

March 2022: Our proposal ‘Using AI to assesss behavior and its relation to internalizing and substance use’ is now funded by UFII Seed Grant award. Co-PI Peter Kvam

January 2022: I accepted the offer to be part of the scientific program committee at CFE CmStatistics. London,Dec 2022.

January 2022: An applied Paper on impact of climate change on campus energy use was accepted at Buildings.

December 2021: We submitted a Paper on forecasting analysis for sources of geopolitical risk.

December 2021: I accepted an invitation for an invited talk at Statfin 2022, ISI Bangalore.

December 2021: Our Paper is accepted at Resources Policy.

December 2021: Our Paper is now accepted at Forecasting. Congrats Kejin on your first paper. This appeared at the Special Feautre edition at the journal selecting some top papers for 2021. We also submitted another paper on similar theme.

November 2021: Our Paper got a revision request from Neural Networks. We submitted the revision.

October 2021: Our Paper on Bitcoin mining activity and volatility dynamics is accepted at Economics Letters. This is a joint work with Rangan Gupta and Riza Demirer.

September 2021: Our Paper on long-term predictions for high-dimensional regression was accepted at Journal of Time-series Analysis.

This was previously prseented in the SWEET pricing invited session at International Symposium of Forecasting, Thessaloniki, Greece June 2019. I am using this video as my submission to the invited poster session of NBER-NSF time-series conference, Rice University, Oct 2021.

August 2021: I was awarded an NSF DMS grant in their ATD program. Award Title: New algorithms for inference and predictions on large geospatial datasets. 2021-24. Role- Solo P.I. $200,000

July 2021: Our Paper on time-series based classifier for HPC based malware detection is now accepted at IEEE International Symposium on Hardware Oriented Security and Trust (HOST) (Acceptance rate 28/130)

June 2021: We submitted a Paper on stochastic neurotron and its convergence guarantee.

June 2021: I was awarded an AMS Simons Travel Grant for 2021-2024.

May 2021: We submitted a paper on Forecasting output growth of Advanced economies where gold market volatility is used as a proxy of global uncertainty. This is a joint work with Aseef Salisu, Rangan Gupta and Sonali Das. Paper

April 2021: Our Paper is now accepted at Electronic Journal of Statistics.

Mar 2021: Our Paper is now accepted at Bayesian Analysis.

Mar 2021: Inducted into Early Career Advisory Board for Journal of Multivariate Analysis. First such experience of being part of an editorial board team!

I also accepted an invitation to serve as a Program Committee Member at CYBER- Sixth International Conference on Cyber-Tenchologies and Cyber-Systems

Feb 2021: Our Paper “Long-term prediction intervals with many covariates” got a revision request from Journal of Time-series Analysis.

Jan 2021: Our Paper “Shrinkage Estimation with Singular Priors and an Application to Small Area Estimation” is now accepted in Journal of Multivariate Analysis.

Jan 2021: We submitted a Paper where we proposed some algorithms with provable guarantees on depth-2 neural nets.

Jan 2021: We submitted a Paper on model-free predictions for econometric datasets. Here we explored a popular model-free prediction based on NoVas transformation for long-horizon prediction and proposed new methods to boost this in diverse scenario.

Dec 2020: We resubmitted Paper on “Regular Stochastic Block Model”

Dec 2020: Our Paper got a revision request from Bayesian Analysis. We submitted the revision.

Nov 2020: Our Paper “Simultaneous inference for time-varying models” is now accepted in Journal of Econometrics.

Oct 2020: Peer-reviewed conference article on cyberbullying gets accepted at HICSS (Hawaii International Conference on System Science) 54. An initial work on change-point exploration was also accepted in 16th Annual social informatics research symposium back in Aug 2020.

Oct 2020: Interesting work has begun on Neural nets convergence, some related new algorithms and their tail properties. We look forward to apply these new algorithms in time-series forecasting.

Sep 2020: We submitted a Paper on time-varying Bayesian ARCH/GARCH model. Here a new B-spline based estimation method is proposed and posterior contraction theory is developed. We were also able to include iGARCH model, which is a first.

Aug 2020: We established posterior contraction theory for time-varying models on Poisson count series. This Paper adds a theoretical proof of the method we proposed in a paper from April analyzing initial Covid-19 counts.

Aug 2020: Our Paper was presented in JSM 2020 in Statistica Sinica invited session. Only 3 published papers in SS between 2018-20 were selected for this distinction.

June 2020: A new project has begun where we focus on spatial patterns in US states Covid-19 counts. In this paper we propose an alternative method of R0 estimation which uses only the time-series data on incidence rate only.

May 2020: A new paper on shrinkage estimators with singular prior is submitted.

April 2020: A Bayesian time-varying model is used to analyze initial Covid-19 propagation across different countries

April 2020: We got a RR from Advances in Applied Probability for our Paper on “Regular Stochastic Block Model”. This is currently being revised.