arXiv subjects

Doudou Zhou

Publications and source records attributed to Doudou Zhou.

At least 19 recordsLinked to original sources

arXiv2026

Conditional Evaluation of Language Models with Cheap Auxiliary Signals

Aggregate accuracy hides where models succeed and fail. Estimating conditional performance profiles from gold labels alone is expensive, while cheap auxiliary signals such as LLM-judge scores, pairwise comparisons, confidence scores, and judge-disagreement features can be collected for every benchmark item but are often biased or miscalibrated. We propose LACE (Local Augmented Control-Variate Evaluation), a semi-supervised estimator for conditional LLM evaluation. The key step is local centering: after subtracting the conditional mean of a cheap signal within the target profile region, any linear augmentation has zero conditional mean and therefore cannot change the estimand. The augmentation coefficient is used only for efficiency, and a local ridge control variate combines a gold-label residual mean from the labeled subset with a cheap-signal mean from the full item pool. We prove calibration-free identification, unbiasedness for grouped profiles, local oracle optimality within centered linear augmentations, and first-order adaptivity to the estimated coefficient. The resulting gain formula is governed by a population local $R^2$, which characterizes how the efficiency attainable from the cheap signals varies across profile values. We also derive corresponding estimators for direct paired model gaps and deployment-weighted scores. We empirically evaluate the primary performance-profile estimator on MATH-500, ScienceQA, MMLU, WinoGrande, HellaSwag, TruthfulQA, GSM8K, and ARC.

cs.LG↗

arXiv2026

Inferential Evaluation of Surrogate-Derived Models under Covariate Shift

In transfer-learning settings, a model derived from abundant surrogate labels may be deployed in a target population where gold-standard outcomes are unobserved. Evaluating its target performance is essential for determining whether decisions based on the model remain reliable, yet it is difficult when gold labels are scarce, and covariate distributions differ across data sources. We study a three-sample setting with a small gold-labeled source, a larger surrogate-labeled source, and an unlabeled target. Under conditional transportability, we evaluate the surrogate-derived model against the latent gold-standard outcome in the target population. We propose cross-fitted estimators that transport information from the two labeled sources through source-specific density ratios. We also combine outcome-regression augmentation with a kernel correction for estimating the model near a threshold, accounting for uncertainty from all three samples. We establish asymptotically linear inference for TPR and FPR, consistency and pointwise inference for the ROC curve, and asymptotically normal inference for AUC. Simulations assess bias, coverage, and sensitivity to bandwidth and relative sample sizes. A retrospective temporal validation on Chatbot Arena and a semi-synthetic ACS-Income study provide validation in real-world AI applications.

stat.ML↗

arXiv2026

Personalizing Large Language Model Agents with Small Policy Models

Large language model (LLM) agents can retrieve memory, call tools, ask clarifying questions, and vary response style, yet adapting these execution decisions to an individual user remains difficult. Fine-tuning a separate LLM is costly or impossible for proprietary systems, while prompts and memory primarily expose user information to the agent rather than adapt its execution decisions from feedback. We formulate personalization of a frozen agent as online learning of a per-user execution policy from scalar feedback observed only for the executed action. We propose FABLE (Factorized Adaptive Bandit Layer for Execution), a lightweight policy layer outside a potentially black-box host agent. FABLE factorizes memory, information-acquisition, and response decisions so feedback updates related choices; filters actions through an externally specified feasible set before exploration; and learns user-specific residual preferences relative to a fixed default-and-cost score via Bayesian contextual Thompson sampling. Under a linear residual-reward model, a calibrated variant inherits an expected-regret bound against the best feasible action. We also characterize preferences unidentifiable under persistent feasibility constraints and provide anytime-valid false-promotion control. Across personalized-reasoning, controlled-feedback, and executable tool-use evaluations, FABLE improves several preference-sensitive behaviors relative to rule-only control while remaining competitive on end-to-end task performance.

cs.AI↗

arXiv2026

Pattern-Calibrated Multimodal Prediction under Blockwise Missingness

Blockwise missingness in multimodal data is usually treated as an incomplete-input problem. We instead focus on prediction for a prespecified observed-modality pattern, where the observed modality set determines the information on which the prediction rule can condition. A procedure that imputes missing modalities, zero-fills unobserved modalities, or trains a single pooled predictor may borrow information across patterns, but it can also mix pattern-specific prediction rules. We propose Multimodal Overlap-aware Shared-specific Alignment and Inter-pattern Calibration (MOSAIC), a pattern-calibrated framework for borrowing across missingness patterns without collapsing their prediction rules. MOSAIC learns shared and modality-specific representations, uses the available representations that overlap with the target pattern to fit a first-stage predictor, and then estimates the calibration gap from target-pattern data. We establish non-asymptotic bounds that decompose the error into overlap effective sample size, calibration gap, and representation-learning error, clarifying when cross-pattern borrowing improves over local fitting and when the improvement is controlled by rule mismatch or representation-learning error. Simulations examine representation recovery and target-pattern correction, and applications to ICU mortality prediction, emotion recognition, and glaucoma classification show gains when target-pattern samples are limited or pattern-specific rules differ.

stat.ME↗

arXiv2026

Structured Transfer Learning for Survival Risk Stratification in Data-Sparse Clinical Cohorts

Background: Survival prediction models are often less reliable in clinical groups with limited sample sizes or few outcome events. Target-only models may be unstable, whereas models from larger cohorts may transfer poorly when risk-factor effects differ across populations. We evaluated whether structured transfer learning can improve survival risk stratification in data-sparse cohorts while allowing cohort-specific adaptation. Methods: We developed the COhort-shared Rank-rEduced Cox model (CORE-Cox), a two-stage framework for multi-outcome survival prediction. CORE-Cox learns shared risk-factor patterns across related outcomes in a larger source cohort via a low-rank Cox coefficient structure, then adapts these patterns to a smaller target cohort through regularized residual correction. We evaluated CORE-Cox in UK Biobank (White source, n=150,093; Asian target, n=2,534) and MIMIC-IV (White ICU source, n=15,997; Asian ICU target, n=672), comparing against target-only Cox, penalized Cox, low-rank multi-task, naive pooling, direct transfer, and single-outcome residual transfer under repeated nested cross-validation. Results: CORE-Cox achieved best or near-best discrimination across most outcomes. Mean C-index improved from 0.733 to 0.766 in UK Biobank and from 0.628 to 0.658 in MIMIC-IV, with gains in eight of nine outcomes. CORE-Cox also improved top-15% risk enrichment, with hazard-ratio estimates typically intermediate between source-only and target-only models. Discussion: CORE-Cox offers an interpretable transfer-learning framework for survival risk stratification in data-sparse cohorts, combining shared cross-outcome structure with cohort-specific adaptation. Further validation is needed before use in calibrated absolute-risk prediction or clinical decision-making.

stat.ME↗

arXiv2026

Spherical Mixture Integration for Latent Embedding Alignment across Multi-Source Feature Spaces

Multi-institutional electronic health record (Multi-EHR) data have emerged as a powerful resource for developing predictive models to support clinical decisions and for generating reliable real-world evidence. By aggregating information from diverse patient populations and institutions, they enhance the robustness and generalizability of models and findings. However, analyzing multi-EHR remains challenging because disparate institutions rarely map all data elements to common ontologies, and raw EHR codes are often overly granular and institution-specific, fragmenting representations of the same clinical concept. Hence, integrative analysis must overcome two key hurdles: harmonizing codes with the same clinical meaning (synonymy), and aligning institutional feature spaces. To address these challenges, we propose SMILE, a Spherical Mixture Integration for Latent Embedding alignment across multi-source feature spaces, where embeddings from heterogeneous sources serve as privacy-preserving summaries of clinical concepts and sparse relational pairs provide weak supervision. Synonymy is modeled via a mixture of von Mises-Fisher distributions, yielding unified representations of semantically equivalent raw codes. We develop a composite quasi-likelihood estimator with non-asymptotic error bounds for the latent representations and mixture mean directions and consistent synonym-cluster recovery, quantifying the gains from integrating multiple sources and knowledge-graph information. Simulations and a multi-institutional EHR application demonstrate improved alignment and synonym clustering.

stat.ME↗

arXiv2026

Unlearning What Matters: Token-Level Attribution for Precise Language Model Unlearning

Machine unlearning has emerged as a critical capability for addressing privacy, safety, and regulatory concerns in large language models (LLMs). Existing methods operate at the sequence level, applying uniform updates across all tokens despite only a subset encoding the knowledge targeted for removal. This introduces gradient noise, degrades utility, and leads to suboptimal forgetting. We propose TokenUnlearn, a token-level attribution framework that identifies and selectively targets critical tokens. Our approach combines knowledge-aware signals via masking, and entropy-aware signals to yield importance scores for precise token selection. We develop two complementary strategies: hard selection, applying unlearning only to high-importance tokens, and soft weighting, modulating gradient contributions based on importance scores. Both extend existing methods to token-level variants. Theoretical analysis shows token-level selection improves gradient signal-to-noise ratio. Experiments on TOFU and WMDP benchmarks across three model architectures demonstrate consistent improvements over sequence-level baselines in both forgetting effectiveness and utility preservation.

cs.CL↗

arXiv2026

Cost-optimal Sequential Testing via Doubly Robust Q-learning

Clinical decision-making often involves selecting tests that are costly, invasive, or time-consuming, motivating individualized, sequential strategies for what to measure and when to stop ascertaining. We study the problem of learning cost-optimal sequential decision policies from retrospective data, where test availability depends on prior results, inducing informative missingness. Under a sequential missing-at-random mechanism, we develop a doubly robust Q-learning framework for estimating optimal policies. The method introduces path-specific inverse probability weights that account for heterogeneous test trajectories and satisfy a normalization property conditional on the observed history. By combining these weights with auxiliary contrast models, we construct orthogonal pseudo-outcomes that enable unbiased policy learning when either the acquisition model or the contrast model is correctly specified. We establish oracle inequalities for the stage-wise contrast estimators, along with convergence rates, regret bounds, and misclassification rates for the learned policy. Simulations demonstrate improved cost-adjusted performance over weighted and complete-case baselines, and an application to a prostate cancer cohort study illustrates how the method reduces testing cost without compromising predictive accuracy.

stat.ML↗

arXiv2026

Hierarchical Contrastive Learning for Multimodal Data

Multimodal representation learning is commonly built on a shared-private decomposition, treating latent information as either common to all modalities or specific to one. This binary view is often inadequate: many factors are shared by only subsets of modalities, and ignoring such partial sharing can over-align unrelated signals and obscure complementary information. We propose Hierarchical Contrastive Learning (HCL), a framework that learns globally shared, partially shared, and modality-specific representations within a unified model. HCL combines a hierarchical latent-variable formulation with structural sparsity and a structure-aware contrastive objective that aligns only modalities that genuinely share a latent factor. Under uncorrelated latent variables, we prove identifiability of the hierarchical decomposition, establish recovery guarantees for the loading matrices, and derive parameter estimation and excess-risk bounds for downstream prediction. Simulations show accurate recovery of hierarchical structure and effective selection of task-relevant components. On multimodal electronic health records, HCL yields more informative representations and consistently improves predictive performance.

stat.ML↗

arXiv2026

Learning Sequential Decisions from Multiple Sources via Group-Robust Markov Decision Processes

We often collect data from multiple sites (e.g., hospitals) that share common structure but also exhibit heterogeneity. This paper aims to learn robust sequential decision-making policies from such offline, multi-site datasets. To model cross-site uncertainty, we study distributionally robust MDPs with a group-linear structure: all sites share a common feature map, and both the transition kernels and expected reward functions are linear in these shared features. We introduce feature-wise (d-rectangular) uncertainty sets, which preserve tractable robust Bellman recursions while maintaining key cross-site structure. Building on this, we then develop an offline algorithm based on pessimistic value iteration that includes: (i) per-site ridge regression for Bellman targets, (ii) feature-wise worst-case (row-wise minimization) aggregation, and (iii) a data-dependent pessimism penalty computed from the diagonals of the inverse design matrices. We further propose a cluster-level extension that pools similar sites to improve sample efficiency, guided by prior knowledge of site similarity. Under a robust partial coverage assumption, we prove a suboptimality bound for the resulting policy. Overall, our framework addresses multi-site learning with heterogeneous data sources and provides a principled approach to robust planning without relying on strong state-action rectangularity assumptions.

stat.ME↗

arXiv2026

A Judge-Aware Ranking Framework for Evaluating Large Language Models without Ground Truth

Evaluating large language models (LLMs) on open-ended tasks without ground-truth labels is increasingly done via the LLM-as-a-judge paradigm. A critical but under-modeled issue is that judge LLMs differ substantially in reliability; treating all judges equally can yield biased leaderboards and misleading uncertainty estimates. More data can make evaluation more confidently wrong under misspecified aggregation. We propose a judge-aware ranking framework that extends the Bradley-Terry-Luce model by introducing judge-specific discrimination parameters, jointly estimating latent model quality and judge reliability from pairwise comparisons without reference labels. We establish identifiability up to natural normalizations and prove consistency and asymptotic normality of the maximum likelihood estimator, enabling confidence intervals for score differences and rank comparisons. Across multiple public benchmarks and a newly collected dataset, our method improves agreement with human preferences, achieves higher data efficiency than unweighted baselines, and produces calibrated uncertainty quantification for LLM rankings.

stat.ML↗

arXiv2026

Preference-based Centrality and Ranking in General Metric Spaces

Ranking or assessing centrality in multivariate and non-Euclidean data is difficult because there is no canonical order and many depth notions become computationally fragile in high-dimensional or structured settings. We introduce a preference-based notion of centrality defined through population proximity comparisons with respect to a random reference draw, yielding a metric-intrinsic statistical functional that is well-defined on general metric spaces. Because the induced pairwise preferences may be non-transitive, we map them to a coherent one-dimensional score via a Bradley--Terry--Luce cross-entropy projection, viewed as a calibrated aggregation device rather than a correctly specified model. We develop two finite-sample estimators a convex M-estimator and a fast spectral estimator based on a comparison operator, and establish identifiability and consistency under mild conditions. Simulations and real-data examples, including high-dimensional and functional observations, illustrate that the proposed scores provide stable, interpretable rankings aligned with the underlying preference centrality.

stat.ME↗

arXiv2026

From Hawkes Processes to Attention: Time-Modulated Mechanisms for Event Sequences

Marked Temporal Point Processes (MTPPs) arise naturally in medical, social, commercial, and financial domains. However, existing Transformer-based methods mostly inject temporal information only via positional encodings, relying on shared or parametric decay structures, which limits their ability to capture heterogeneous and type-specific temporal effects. Inspired by this observation, we derive a novel attention operator called Hawkes Attention from the multivariate Hawkes process theory for MTPP, using learnable per-type neural kernels to modulate query, key and value projections, thereby replacing the corresponding parts in the traditional attention. Benefited from the design, Hawkes Attention unifies event timing and content interaction, learning both the time-relevant behavior and type-specific excitation patterns from the data. The experimental results show that our method achieves better performance compared to the baselines. In addition to the general MTPP, our attention mechanism can also be easily applied to specific temporal structures, such as time series forecasting.

cs.LG↗

arXiv2025

A Trainable Centrality Framework for Modern Data

Measuring how central or typical a data point is underpins robust estimation, ranking, and outlier detection, but classical depth notions become expensive and unstable in high dimensions and are hard to extend beyond Euclidean data. We introduce Fused Unified centrality Score Estimation (FUSE), a neural centrality framework that operates on top of arbitrary representations. FUSE combines a global head, trained from pairwise distance-based comparisons to learn an anchor-free centrality score, with a local head, trained by denoising score matching to approximate a smoothed log-density potential. A single parameter between 0 and 1 interpolates between these calibrated signals, yielding depth-like centrality from different views via one forward pass. Across synthetic distributions, real images, time series, and text data, and standard outlier detection benchmarks, FUSE recovers meaningful classical ordering, reveals multi-scale geometric structures, and attains competitive performance with strong classical baselines while remaining simple and efficient.

cs.LG↗

arXiv2025

RELEAP: Reinforcement-Enhanced Label-Efficient Active Phenotyping for Electronic Health Records

Objective: Electronic health record (EHR) phenotyping often relies on noisy proxy labels, which undermine the reliability of downstream risk prediction. Active learning can reduce annotation costs, but most rely on fixed heuristics and do not ensure that phenotype refinement improves prediction performance. Our goal was to develop a framework that directly uses downstream prediction performance as feedback to guide phenotype correction and sample selection under constrained labeling budgets. Materials and Methods: We propose Reinforcement-Enhanced Label-Efficient Active Phenotyping (RELEAP), a reinforcement learning-based active learning framework. RELEAP adaptively integrates multiple querying strategies and, unlike prior methods, updates its policy based on feedback from downstream models. We evaluated RELEAP on a de-identified Duke University Health System (DUHS) cohort (2014-2024) for incident lung cancer risk prediction, using logistic regression and penalized Cox survival models. Performance was benchmarked against noisy-label baselines and single-strategy active learning. Results: RELEAP consistently outperformed all baselines. Logistic AUC increased from 0.774 to 0.805 and survival C-index from 0.718 to 0.752. Using downstream performance as feedback, RELEAP produced smoother and more stable gains than heuristic methods under the same labeling budget. Discussion: By linking phenotype refinement to prediction outcomes, RELEAP learns which samples most improve downstream discrimination and calibration, offering a more principled alternative to fixed active learning rules. Conclusion: RELEAP optimizes phenotype correction through downstream feedback, offering a scalable, label-efficient paradigm that reduces manual chart review and enhances the reliability of EHR-based risk prediction.

cs.LG↗

arXiv2025

A Common Pipeline for Harmonizing Electronic Health Record Data for Translational Research

Despite the growing availability of Electronic Health Record (EHR) data, researchers often face substantial barriers in effectively using these data for translational research due to their complexity, heterogeneity, and lack of standardized tools and documentation. To address this critical gap, we introduce PEHRT, a common pipeline for harmonizing EHR data for translational research. PEHRT is a comprehensive, ready-to-use resource that includes open-source code, visualization tools, and detailed documentation to streamline the process of preparing EHR data for analysis. The pipeline provides tools to harmonize structured and unstructured EHR data to standardized ontologies to ensure consistency across diverse coding systems. In the presence of unmapped or heterogeneous local codes, PEHRT further leverages representation learning and pre-trained language models to generate robust embeddings that capture semantic relationships across sites to mitigate heterogeneity and enable integrative downstream analyses. PEHRT also supports cross-institutional co-training through shared representations, allowing participating sites to collaboratively refine embeddings and enhance generalizability without sharing individual-level data. The framework is data model-agnostic and can be seamlessly deployed across diverse healthcare systems to produce interoperable, research-ready datasets. By lowering the technical barriers to EHR-based research, PEHRT empowers investigators to transform raw clinical data into reproducible, analysis-ready resources for discovery and innovation.

stat.ML↗

arXiv2025

Automated Hierarchical Graph Construction for Multi-source Electronic Health Records

Electronic Health Records (EHRs), comprising diverse clinical data such as diagnoses, medications, and laboratory results, hold great promise for translational research. EHR-derived data have advanced disease prevention, improved clinical trial recruitment, and generated real-world evidence. Synthesizing EHRs across institutions enables large-scale, generalizable studies that capture rare diseases and population diversity, but remains hindered by the heterogeneity of medical codes, institution-specific terminologies, and the absence of standardized data structures. These barriers limit the interpretability, comparability, and scalability of EHR-based analyses, underscoring the need for robust methods to harmonize and extract meaningful insights from distributed, heterogeneous data. To address this, we propose MASH (Multi-source Automated Structured Hierarchy), a fully automated framework that aligns medical codes across institutions using neural optimal transport and constructs hierarchical graphs with learned hyperbolic embeddings. During training, MASH integrates information from pre-trained language models, co-occurrence patterns, textual descriptions, and supervised labels to capture semantic and hierarchical relationships among medical concepts more effectively. Applied to real-world EHR data, including diagnosis, medication, and laboratory codes, MASH produces interpretable hierarchical graphs that facilitate the navigation and understanding of heterogeneous clinical data. Notably, it generates the first automated hierarchies for unstructured local laboratory codes, establishing foundational references for downstream applications.

stat.ML↗

arXiv2025

WISE: A Weighted Similarity Aggregation Test for Serial Independence

We propose a nonparametric test for serial independence that aggregates pairwise similarities of observations with lag-dependent weights. The resulting statistic is powerful to general forms of temporal dependence, including nonlinear and uncorrelated alternatives, and applies to ultra-high-dimensional and non-Euclidean data. We derive asymptotic normality under both permutation and population nulls, and establish consistency in classical large-sample and high-dimension-low-sample-size (HDLSS) regimes. The test therefore provides the first theoretical power guarantees for serial independence in the HDLSS setting. Simulations demonstrate accurate size and strong power against a wide range of alternatives, showing significant power improvement over existing methods under various high-dimensional time series models. An application to spatio-temporal data illustrates the method's utility for non-Euclidean observations.

stat.ME↗