11 papers
Byzantine-Robust Federated RAG via Aligned Calibration and Fixed-Membership Conformal Prediction
Prasanjit Dubey, Xiaoming Huo, Aritra Guha
Retrieval-augmented generation (RAG) lets language models answer questions more accurately by consulting relevant documents. Many valuable collections, such as medical records, can…
Pooling Sequential Evidence Across Hypotheses: Rate-Optimal Multiple Testing at a Fixed Horizon
Prasanjit Dubey, Xiaoming Huo
We study sequential testing of a fixed family of hypotheses when observations are costly and a sampling horizon is specified in advance. The challenge is to pool evidence for earli…
Federated multiple testing with quantized p-values under family-wise error control
Prasanjit Dubey, Xiaoming Huo
Several institutions test one family of hypotheses under family-wise error control but cannot pool their data, so each site releases only a quantized report of its p-value for each…
Matching Rates and Optimal Allocation for Federated Probe-Logit Distillation under Heterogeneous Bandwidth Budgets
Prasanjit Dubey, Xiaoming Huo
In federated language modeling, nodes each hold samples but cannot pool data or exchange full-precision gradients or weights. We study the minimax rate at which a condition…
Certified Adaptive Refresh: Anytime-Valid Monitoring for Federated Conformal RAG
Prasanjit Dubey, Xiaoming Huo
Question-answering services built on retrieval-augmented generation (RAG), in which a language model answers from retrieved documents, are inspected continuously and upgraded repea…
BOOST: Power-Optimal Strong-FWER Testing for Block-Structured Multiplicity
Prasanjit Dubey, Xiaoming Huo
Structured multiple-testing problems (gatekeeping trials, dose-finding, multi-tissue eQTL mapping, bundled-challenger A/B experiments) organize hypotheses into design-imposed block…
Federated Language Models Under Bandwidth Budgets: Distillation Rates and Conformal Coverage
Prasanjit Dubey, Xiaoming Huo
Hospitals, companies, and research consortia want language models that learn from, and answer questions over, data that must stay where it is held, over links with limited bandwidt…
Optimal multiple testing under family-wise error control: elementary symmetric polynomials and a scalable algorithm
Prasanjit Dubey, Xiaoming Huo
Family-wise error rate control is essential when even one false rejection is costly, but distribution-free procedures do not exploit information in a specified alternative model. E…
LLmFPCA-detect: LLM-powered Multivariate Functional PCA for Anomaly Detection in Sparse Longitudinal Texts
Prasanjit Dubey, Aritra Guha, Zhengyi Zhou +3
Sparse longitudinal (SL) textual data arises when individuals generate text repeatedly over time (e.g., customer reviews, occasional social media posts, electronic medical records…
Most Powerful Test with Exact Family-Wise Error Rate Control: Necessary Conditions and a Path to Fast Computing
Prasanjit Dubey, Xiaoming Huo
Identifying the most powerful test in multiple hypothesis testing under strong family-wise error rate (FWER) control is a fundamental problem in statistical methodology. State-of-t…
Generalization Bounds for Magnitude-Based Pruning via Sparse Matrix Sketching
Etash Kumar Guha, Prasanjit Dubey, Xiaoming Huo
In this paper, we derive a novel bound on the generalization error of Magnitude-Based pruning of overparameterized neural networks. Our work builds on the bounds in Arora et al. [2…