About

Jyotishka Datta is an Associate Professor of Statistics at Virginia Tech, where he previously served as an Assistant Professor. Prior to joining Virginia Tech, he was an Assistant Professor in the Department of Mathematical Sciences at the University of Arkansas, Fayetteville from 2016 to 2020. His research focuses on Bayesian methodology and theory for structured high-dimensional data, including areas such as shrinkage estimation, sparse signal recovery, graphical modeling, and nonparametric Bayes. His work has applications across diverse fields including astronomy, cancer genomics, neuroscience, ecology, and crime forecasting. Dr. Datta received his Ph.D. in Statistics from Purdue University under the guidance of Professors Jayanta K. Ghosh and Michael Yu Zhu, with a thesis on theoretical and methodological aspects of multiple testing and model selection. He also completed postdoctoral training at Duke University and SAMSI, working with Professors David B. Dunson and Sandeep S. Dave. Among his honors, he was awarded the NSF CAREER Award in 2025 and the Dayanand Naik Award from the Virginia Chapter of the American Statistical Association in 2023, recognizing his outstanding research contributions and service in statistics.

Research topics

• Computer Science
• Medicine
• Artificial Intelligence
• Machine Learning
• Virology
• Internal medicine
• Statistics
• Pathology
• Meteorology
• Mathematics
• Geography
• Pediatrics

Selected publications

• A New Look at Bayesian Testing

  ArXiv.org · 2026-02-11

  articleOpen access1st authorCorresponding

  We identify the critical deviation scale governing Bayesian evidence accumulation in regular parametric testing. Under integrated Bayes risk with zero-one loss, the risk-optimal rejection boundary lies in a moderate deviation regime, with a square-root logarithmic inflation relative to the usual local asymptotic normal scale. Under Cramer regularity, local prior smoothness at the null, and symmetric loss, we derive the sharp threshold and show that its leading logarithmic term is universal across regular priors, while lower-order constants depend on the local prior density, Fisher information, and prior model odds. The result extends to one-parameter exponential families through local asymptotic normality and places Jeffreys' testing threshold, the Bayesian information criterion penalty, and Chernoff-Stein type error-exponent arguments within a common asymptotic moderate deviation framework.

  PublisherOA PDF

• Inverse Probability Weighting: From Survey Sampling to Evidence Estimation

  The New England Journal of Statistics in Data Science · 2026-01-01

  preprintOpen access1st authorCorresponding

  We consider the class of inverse probability weight (IPW) estimators, including the popular Horvitz–Thompson and Hájek estimators used routinely in survey sampling, causal inference and for Bayesian computation. We focus on the ‘weak paradoxes’ for these estimators due to two counterexamples by Basu (1988) and Wasserman (2004) and investigate the two natural Bayesian answers to this problem: one based on binning and smoothing: a ‘Bayesian sieve’ and the other based on a conjugate hierarchical model that allows borrowing information via exchangeability. We compare the mean squared errors for the two Bayesian estimators with the IPW estimators for Wasserman’s example via simulation studies on a broad range of parameter configurations. We also prove posterior consistency for the Bayes estimators under missing-completely-at-random assumption and show that it requires fewer assumptions on the inclusion probabilities. We also revisit the connection between the different problems where improved or adaptive IPW estimators will be useful, including survey sampling, evidence estimation strategies such as Conditional Monte Carlo, Riemannian sum, Trapezoidal rules and vertical likelihood, as well as average treatment effect estimation in causal inference.

  PublisherOA PDFDOI

• A New Look at Bayesian Testing

  Open MIND · 2026-02-11

  preprint1st authorCorresponding

  We identify the critical deviation scale governing Bayesian evidence accumulation in regular parametric testing. Under integrated Bayes risk with zero-one loss, the risk-optimal rejection boundary lies in a moderate deviation regime, with a square-root logarithmic inflation relative to the usual local asymptotic normal scale. Under Cramer regularity, local prior smoothness at the null, and symmetric loss, we derive the sharp threshold and show that its leading logarithmic term is universal across regular priors, while lower-order constants depend on the local prior density, Fisher information, and prior model odds. The result extends to one-parameter exponential families through local asymptotic normality and places Jeffreys' testing threshold, the Bayesian information criterion penalty, and Chernoff-Stein type error-exponent arguments within a common asymptotic moderate deviation framework.

  DOI

• Polynomial Log-Marginals and Tweedie's Formula : When Is Bayes Possible?

  ArXiv.org · 2025-09-06

  preprintOpen access1st authorCorresponding

  Motivated by Tweedie's formula for the Compound Decision problem, we examine the theoretical foundations of empirical Bayes estimators that directly model the marginal density $m(y)$. Our main result shows that polynomial log-marginals of degree $k \ge 3 $ cannot arise from any valid prior distribution in exponential family models, while quadratic forms correspond exactly to Gaussian priors. This provides theoretical justification for why certain empirical Bayes decision rules, while practically useful, do not correspond to any formal Bayes procedures. We also strengthen the diagnostic by showing that a marginal is a Gaussian convolution only if it extends to a bounded solution of the heat equation in a neighborhood of the smoothing parameter, beyond the convexity of $c(y)=\tfrac12 y^2+\log m(y)$.

  PublisherOA PDFDOI

• Conformal Prediction = Bayes?

  ArXiv.org · 2025-12-29

  articleOpen access1st authorCorresponding

  Conformal prediction (CP) is widely presented as distribution-free predictive inference with finite-sample marginal coverage under exchangeability. We argue that CP is best understood as a rank-calibrated descendant of the Fisher-Dempster-Hill fiducial/direct-probability tradition rather than as Bayesian conditioning in disguise. We establish four separations from coherent countably additive predictive semantics. First, canonical conformal constructions violate conditional extensionality: prediction sets can depend on the marginal design P(X) even when P(Y|X) is fixed. Second, any finitely additive sequential extension preserving rank calibration is nonconglomerable, implying countable Dutch-book vulnerabilities. Third, rank-calibrated updates cannot be realized as regular conditionals of any countably additive exchangeable law on Y^infty. Fourth, formalizing both paradigms as families of one-step predictive kernels, conformal and Bayesian kernels coincide only on a Baire-meagre subset of the space of predictive laws. We further show that rank- and proxy-based reductions are generically Blackwell-deficient relative to full-data experiments, yielding positive Le Cam deficiency for suitable losses. Extending the analysis to prediction-powered inference (PPI) yields an analogous message: bias-corrected, proxy-rectified estimators can be valid as confidence devices while failing to define transportable belief states across stages, shifts, or adaptive selection. Together, the results sharpen a general limitation of wrappers: finite-sample calibration guarantees do not by themselves supply composable semantics for sequential updating or downstream decision-making.

  PublisherOA PDF

• Conformal Prediction = Bayes?

  arXiv (Cornell University) · 2025-12-29

  preprintOpen access1st authorCorresponding

  Conformal prediction (CP) is widely presented as distribution-free predictive inference with finite-sample marginal coverage under exchangeability. We argue that CP is best understood as a rank-calibrated descendant of the Fisher-Dempster-Hill fiducial/direct-probability tradition rather than as Bayesian conditioning in disguise. We establish four separations from coherent countably additive predictive semantics. First, canonical conformal constructions violate conditional extensionality: prediction sets can depend on the marginal design P(X) even when P(Y|X) is fixed. Second, any finitely additive sequential extension preserving rank calibration is nonconglomerable, implying countable Dutch-book vulnerabilities. Third, rank-calibrated updates cannot be realized as regular conditionals of any countably additive exchangeable law on Y^infty. Fourth, formalizing both paradigms as families of one-step predictive kernels, conformal and Bayesian kernels coincide only on a Baire-meagre subset of the space of predictive laws. We further show that rank- and proxy-based reductions are generically Blackwell-deficient relative to full-data experiments, yielding positive Le Cam deficiency for suitable losses. Extending the analysis to prediction-powered inference (PPI) yields an analogous message: bias-corrected, proxy-rectified estimators can be valid as confidence devices while failing to define transportable belief states across stages, shifts, or adaptive selection. Together, the results sharpen a general limitation of wrappers: finite-sample calibration guarantees do not by themselves supply composable semantics for sequential updating or downstream decision-making.

  PublisherDOI

• Increased listening effort and cochlear neural degeneration underlie speech-in-noise deficits in normal-hearing middle-aged adults

  eLife · 2025-07-22

  articleOpen access

  Middle age represents a critical period of accelerated brain changes and provides a window for early detection and intervention in age-related neurological decline. Hearing loss is a key early marker of such decline and is linked to numerous comorbidities in older adults. Yet, ~10% of middle-aged individuals who report hearing difficulties show normal audiograms. Cochlear neural degeneration (CND) could contribute to these hidden hearing deficits, though its role remains unclear due to a lack of objective diagnostics and uncertainty regarding its perceptual outcomes. Here, we employed a cross-species design to examine neural and behavioral signatures of CND. We measured envelope following responses (EFRs) – neural ensemble responses to sound originating from the peripheral auditory pathway – in young and middle-aged adults with normal audiograms and compared these responses to young and middle-aged Mongolian gerbils, where CND was histologically confirmed. We observed near-identical changes in EFRs across species that were associated with CND. Behavioral assessments revealed age-related speech-in-noise deficits under challenging conditions, while pupil-indexed listening effort increased with age even when behavioral performance was matched. Together, these results demonstrate that CND contributes to speech perception difficulties and elevated listening effort in midlife, which may ultimately lead to listening fatigue and social withdrawal.

  PublisherDOI

• Quantile importance sampling

  Brazilian Journal of Probability and Statistics · 2025-09-01

  articleOpen access1st authorCorresponding

  In Bayesian inference, the approximation of integrals of the form ψ=EFl(X)=∫χl(x)dF(x) is a fundamental challenge. Such integrals are crucial for evidence estimation, which is important for various purposes, including model selection and numerical analysis. The existing strategies for evidence estimation are classified into four categories: deterministic approximation, density estimation, importance sampling and vertical representation (SIAM Review 65 (2023) 3–58). In this paper, we show that the Riemann sum estimator due to (SIAM Journal on Numerical Analysis 15 (1978) 1289–1300) can be used in the context of nested sampling (Bayesian Analysis 1 (2006) 833–859) to achieve a O(n−4) rate of convergence, faster than the usual ergodic central limit theorem, under certain regularity conditions. We provide a brief overview of the literature on the Riemann sum estimators and the nested sampling algorithm and its connections to vertical likelihood Monte Carlo. We provide theoretical and numerical arguments to show how merging these two ideas may result in improved and more robust estimators for evidence estimation, especially in higher-dimensional spaces. We also briefly discuss the idea of simulating the Lorenz curve that avoids the problem of intractable Λ functions, essential for the vertical representation and nested sampling.

  PublisherOA PDFDOI

• Author response: Increased listening effort and cochlear neural degeneration underlie behavioral deficits in speech perception in noise in normal hearing middle-aged adults

  2025-06-26

  peer-reviewOpen access

  Middle-age is a critical period of rapid changes in brain function that presents an opportunity for early diagnostics and intervention for neurodegenerative conditions later in life. Hearing loss is one such early indicator linked to many comorbidities experienced in older age. However, current clinical tests fail to capture hearing difficulties for ∼10% of middle-aged adults with normal hearing thresholds seeking help at hearing clinics. Cochlear neural degeneration (CND) could play a role in these hearing deficits, but our current understanding is limited by the lack of objective diagnostics and uncertainty regarding its perceptual consequences. Here, using a cross-species approach, we measured envelope following responses (EFRs) – neural ensemble responses to sound originating from the peripheral auditory pathway – in young and middle-aged adults with normal audiometric thresholds and compared these responses to young and middle-aged Mongolian gerbils, where CND was histologically confirmed. We observed near identical changes in EFRs across species that were associated with CND. Perceptual effects measured as behavioral readouts showed deficits in the most challenging listening conditions and were associated with CND. Additionally, pupil-indexed listening effort increased even at moderate task difficulties where behavioral outcomes were matched. Our results reveal perceptual deficits in middle-aged adults are associated with CND and increases in listening effort, which may result in increased listening fatigue and conversational disengagement.

  PublisherDOI

• Bayesian Global-Local Regularization

  ArXiv.org · 2025-12-16

  preprintOpen access1st authorCorresponding

  We propose a unified framework for global-local regularization that bridges the gap between classical techniques -- such as ridge regression and the nonnegative garotte -- and modern Bayesian hierarchical modeling. By estimating local regularization strengths via marginal likelihood under order constraints, our approach generalizes Stein's positive-part estimator and provides a principled mechanism for adaptive shrinkage in high-dimensional settings. We establish that this isotonic empirical Bayes estimator achieves near-minimax risk (up to logarithmic factors) over sparse ordered model classes, constituting a significant advance in high-dimensional statistical inference. Applications to orthogonal polynomial regression demonstrate the methodology's flexibility, while our theoretical results clarify the connections between empirical Bayes, shape-constrained estimation, and degrees-of-freedom adjustments.

  PublisherOA PDFDOI

Frequent coauthors

• Anindya Bhadra

  Purdue University West Lafayette

  45 shared

• Nicholas G. Polson

  39 shared

• Brandon T. Willard

  34 shared

• Sandeep S. Davé

  University College London

  16 shared

• Rex Au-Yeung

  University of Hong Kong

  14 shared

• Cassandra Love

  Duke University

  14 shared

• Deepthi Rajagopalan

  12 shared

• Magdalena Czader

  Indiana University School of Medicine

  12 shared

Education

• Other

  Indian Statistical Institute

  2008

• Other

  Indian Statistical Institute

  2003

• Ph.D., Statistics

  Purdue University

  2014

Awards & honors

• NSF CAREER Award (2025)
• Dayanand Naik Award from the Virginia Chapter of American St…
• Robert and Sandra Connor Endowed Faculty Fellowship from the…
• Honorable Mention Award for Best Theoretical Poster at the O…
• William J. Studden Publication Award for an outstanding publ…