arXiv subjects

Dehan Kong

Publications and source records attributed to Dehan Kong.

At least 19 recordsLinked to original sources

arXiv2026

A Unified Rate-Distortion Perspective on Vector, Product, and Scalar Quantization

Discrete visual tokenization, predominantly driven by vector, scalar, and product quantization, lacks a unified conceptual framework for understanding quantization tradeoffs. In this paper, we propose a unified rate--distortion perspective on modern discrete visual tokenization. By viewing quantization as lossy compression, we characterize the nominal fixed-length coding rate through token count and codebook size, and quantization error as the distortion. Within this framework, we resolve three central questions. First, we theoretically and empirically show that minimizing distortion, rather than maximizing codebook utilization, is the primary intrinsic objective for reconstruction fidelity, with a direct connection to the STE-induced gradient discrepancy. Second, we establish two critical fairness conditions for intrinsic quantization comparison: controlling latent feature statistics and enforcing identical coding rates. Third, under these conditions, we recover the VQ--PQ--SQ distortion hierarchy in modern visual tokenization and show empirically that modern VQ methods achieve the lowest distortion. This work provides a foundational rate--distortion reframing of modern discrete visual tokenization, resolves ambiguities in quantizer evaluation, and provides a controlled framework for isolating intrinsic quantization effectiveness under fixed-rate constraints.

cs.LG↗

arXiv2026

VQ-Transplant: Efficient VQ-Module Integration for Pre-trained Visual Tokenizers

Vector Quantization (VQ) underpins modern discrete visual tokenization. However, training quantization modules for state-of-the-art VQ-based models requires significant computational resources which, in practice, all but prevents the development of novel, cutting-edge VQ techniques under resource constraints. To address this limitation, we propose {\bf VQ-Transplant}, a simple framework that enables plug-and-play integration of new VQ modules into frozen, pre-trained tokenizers by replacing their native VQ modules. Crucially, the proposed transplantation process preserves all encoder-decoder parameters, obviating the need for costly end-to-end retraining when modifying the quantization method. To mitigate decoder-quantization mismatch, we introduce a lightweight decoder adaptation strategy (trained for only 5 epochs on ImageNet-1k) to align feature priors with the new quantization space. In our empirical evaluation, we find that VQ-Transplant allows obtaining near state-of-the-art reconstruction fidelity for industry-level models like VAR while reducing the training cost by 95\%. VQ-Transplant democratizes quantization research by enabling resource-efficient integration of novel VQ techniques while matching industry-level reconstruction performance.

cs.CV↗

arXiv2026

A Statistically Reliable Optimization Framework for Bandit Experiments in Scientific Discovery

Scientific experimentation is largely driven by statistical hypothesis testing to determine significant differences in interventions. Traditionally, experimenters allocate samples uniformly between each intervention. However, such an approach may lead to suboptimal outcomes - multi-armed bandits (MABs) addresses this problem by allocating samples adaptively to maximize outcomes. Yet, two challenges have hindered the use of MABs in scientific domains. First, common hypothesis tests (e.g., $t$-tests) become invalid under adaptive sampling without correction, leading to inflated type~I and type~II errors. This is an understudied problem, and prior solutions suffer from issues such as low statistical power which prevent adoption in many practical settings. Second, practitioners must explicitly balance cumulative reward with statistical efficiency, yet no general methodology exists to quantify this trade-off across algorithms. In this paper, we study assumption modification and critical region correction approaches for hypothesis testing that enable common tests to be applied to adaptively collected data. We provide heuristic justification for its power efficiency and show in simulation that it achieves higher power than existing approaches. Further, we derive a theoretically and practically motivated objective function for adaptive experiment evaluation, which we integrate into a unified experimental framework. Our framework asks experimenters to specify an experiment extension cost for their problem, and based on that enables our proposed optimization procedure to select the bandit algorithm that best balances reward and power in their setting. We show that our approach enables practitioners to improve outcomes with only slightly more steps than uniform randomization, while retaining statistical validity.

stat.AP↗

arXiv2026

RIE-Greedy: Regularization-Induced Exploration for Contextual Bandits

Real-world contextual bandit problems with complex reward models are often tackled with iteratively trained models, such as boosting trees. However, it is difficult to directly apply simple and effective exploration strategies--such as Thompson Sampling or UCB--on top of those black-box estimators. Existing approaches rely on sophisticated assumptions or intractable procedures that are hard to verify and implement in practice. In this work, we explore the use of an exploration-free (pure-greedy) action selection strategy, that exploits the randomness inherent in model fitting process as an intrinsic source of exploration. More specifically, we note that the stochasticity in cross-validation based regularization process can naturally induce Thompson Sampling-like exploration. We show that this regularization-induced exploration is theoretically equivalent to Thompson Sampling in the two-armed bandit case and empirically leads to reliable exploration in large-scale business environments compared to benchmark methods such as epsilon-greedy and other state-of-the-art approaches. Overall, our work reveals how regularized estimator training itself can induce effective exploration, offering both theoretical insight and practical guidance for contextual bandit design.

stat.ML↗

arXiv2026

WebFactory: Automated Compression of Foundational Language Intelligence into Grounded Web Agents

Current paradigms for training GUI agents are fundamentally limited by a reliance on either unsafe, non-reproducible live web interactions or costly, scarce human-crafted data and environments. We argue this focus on data volume overlooks a more critical factor: the efficiency of compressing a large language model's (LLM) latent knowledge into actionable agent behavior. We introduce WebFactory, a novel, fully automated closed-loop reinforcement learning pipeline for GUI agents, systematically compressing LLM-encoded internet intelligence into efficient, grounded actions. Our pipeline features a process of scalable environment synthesis, knowledge-aware task generation, LLM-powered trajectory collection, decomposed reward RL training, and systematic agent evaluation. Remarkably, our agent demonstrates exceptional data efficiency and generalization. Trained on synthetic data from only 10 websites within WebFactory, it achieves performance comparable to GUI agents trained on the same amount of human-annotated data from a much larger set of environments. This superior performance is consistent across our internal offline and online transfer benchmarks, where our agent also significantly outperforms the base foundation model. We further provide critical insights into the "embodiment potential" of different LLM foundations, offering a new axis for model evaluation. This work presents a scalable and cost-effective paradigm for transforming passive internet knowledge into active, grounded intelligence, marking a critical step towards general-purpose interactive agents.

cs.AI↗

arXiv2026

WebChain: A Large-Scale Human-Annotated Dataset of Real-World Web Interaction Traces

We introduce WebChain, the largest open-source dataset of human-annotated trajectories on real-world websites, designed to accelerate reproducible research in web agents. It contains 31,725 trajectories and 318k steps, featuring a core Triple Alignment of visual, structural, and action data to provide rich, multi-modal supervision. The data is collected via a scalable pipeline that ensures coverage of complex, high-value tasks often missed by synthetic methods. Leveraging this dataset, we propose a Dual Mid-Training recipe that decouples spatial grounding from planning, achieving state-of-the-art performance on our proposed WebChainBench and other public GUI benchmarks. Our work provides the data and insights necessary to build and rigorously evaluate the next generation of scalable web agents.

cs.AI↗

arXiv2025

VenusBench-GD: A Comprehensive Multi-Platform GUI Benchmark for Diverse Grounding Tasks

GUI grounding is a critical component in building capable GUI agents. However, existing grounding benchmarks suffer from significant limitations: they either provide insufficient data volume and narrow domain coverage, or focus excessively on a single platform and require highly specialized domain knowledge. In this work, we present VenusBench-GD, a comprehensive, bilingual benchmark for GUI grounding that spans multiple platforms, enabling hierarchical evaluation for real-word applications. VenusBench-GD contributes as follows: (i) we introduce a large-scale, cross-platform benchmark with extensive coverage of applications, diverse UI elements, and rich annotated data, (ii) we establish a high-quality data construction pipeline for grounding tasks, achieving higher annotation accuracy than existing benchmarks, and (iii) we extend the scope of element grounding by proposing a hierarchical task taxonomy that divides grounding into basic and advanced categories, encompassing six distinct subtasks designed to evaluate models from complementary perspectives. Our experimental findings reveal critical insights: general-purpose multimodal models now match or even surpass specialized GUI models on basic grounding tasks. In contrast, advanced tasks, still favor GUI-specialized models, though they exhibit significant overfitting and poor robustness. These results underscore the necessity of comprehensive, multi-tiered evaluation frameworks.

cs.CV↗

arXiv2025

Convergence and Optimality of the EM Algorithm Under Multi-Component Gaussian Mixture Models

Gaussian mixture models (GMMs) are fundamental statistical tools for modeling heterogeneous data. Due to the nonconcavity of the likelihood function, the Expectation-Maximization (EM) algorithm is widely used for parameter estimation of each Gaussian component. Existing analyses of the EM algorithm's convergence to the true parameter focus on either the two-component case or multi-component settings with known mixing probabilities and isotropic covariance matrices. In this work, we study the convergence of the EM algorithm for multi-component GMMs in full generality. The population-level EM is shown to converge to the true parameter when the smallest separation among all pairs of Gaussian components exceeds a logarithmic factor of the largest separation and the reciprocal of the minimal mixing probabilities. At the sample level, the EM algorithm is shown to be minimax rate-optimal, up to a logarithmic factor. We develop two distinct novel analytical approaches, each tailored to a different regime of separation, reflecting two complementary perspectives on the use of EM. As a byproduct of our analysis, we show that the EM algorithm, when used for community detection, also achieves the minimax optimal rate of misclustering error under milder separation conditions than spectral clustering and Lloyd's algorithm, an interesting result in its own right. Our analysis allows the number of components, the minimal mixing probabilities, the separation between Gaussian components and the dimension to grow with the sample size. Simulation studies corroborate our theoretical findings.

math.ST↗

arXiv2025

Successive classification learning for estimating quantile optimal treatment regimes

Quantile optimal treatment regimes (OTRs) aim to assign treatments that maximize a specified quantile of patients' outcomes. Compared to treatment regimes that target the mean outcomes, quantile OTRs offer fairer regimes when a lower quantile is selected, as it improves outcomes for vulnerable patients. In this paper, we propose a novel method for estimating quantile OTRs by reformulating the problem as a successive classification task, solvable via training a sequence of classifiers, each successive classifier built on the output of its predecessors. This reformulation enables us to leverage the powerful machine learning technique to enhance computational efficiency and handle complex decision boundaries. We also investigate the estimation of quantile OTRs when outcomes are discrete, a setting that has received limited attention in the literature. A key challenge is that direct extensions of existing methods to discrete outcomes often lead to inconsistency and ineffectiveness issues. To overcome this, we introduce a smoothing technique that maps discrete outcomes to continuous surrogates, enabling consistent and effective estimation. We provide theoretical guarantees to support our methodology, and demonstrate its superior performance through comprehensive simulation studies and real-data analysis.

stat.ME↗

arXiv2025

Universal Retrieval for Multimodal Trajectory Modeling

Trajectory data, capturing human actions and environmental states across various modalities, holds significant potential for enhancing AI agent capabilities, particularly in GUI environments. However, how to model the representation of trajectory-level data presents a significant challenge that has not been systematically addressed amid explosive trajectory data growth. In this work, we introduce Multimodal Trajectory Retrieval, bridging the gap between universal retrieval and agent-centric trajectory modeling. We construct the Unified Agent Trajectory Dataset (UATD) from annotated demonstrations and states across diverse real-world scenarios. Based on this, we present GAE-Bench, a benchmark containing a large number of trajectory-based retrieval pairs. In addition, we propose GAE-Retriever, a multimodal retrieval framework that adopts vision-language models and incorporates optimized contrastive learning through a token selection and the GradCache mechanism. Comprehensive evaluations across multiple datasets show that GAE-Retriever consistently outperforms strong baselines in retrieval recall, highlighting its effectiveness in advancing multimodal trajectory retrieval.

cs.AI↗

arXiv2025

Simultaneous hypothesis testing for comparing many functional means

Data with multiple functional recordings at each observational unit are increasingly common in various fields including medical imaging and environmental sciences. To conduct inference for such observations, we develop a paired two-sample test that allows to simultaneously compare the means of many functional observations while maintaining family-wise error rate control. We explicitly allow the number of functional recordings to increase, potentially much faster than the sample size. Our test is fully functional and does not rely on dimension reduction or functional PCA type approaches or the choice of tuning parameters. To provide a theoretical justification for the proposed procedure, we develop a number of new anti-concentration and Gaussian approximation results for maxima of $L^2$ statistics which might be of independent interest. The methodology is illustrated on the task-related cortical surface functional magnetic resonance imaging data from Human Connectome Project.

stat.ME↗

arXiv2025

Towards R-learner with Continuous Treatments

The R-learner is widely used in causal inference due to its flexibility and efficiency in estimating the conditional average treatment effect. However, extending the R-learner framework from binary to continuous treatments introduces a non-identifiability issue, as the functional zero constraint inherent to the conditional average treatment effect cannot be directly imposed in the R-loss under continuous treatments. To address this, we propose a two-step identification strategy: we first identify an intermediary function via Tikhonov regularization, and then recover the conditional average treatment effect using a zero-constraining operator. Building on this strategy, an $\ell_2$-regularized R-learner framework is developed to estimate the conditional average treatment effect for continuous treatments. The new framework accommodates modern, flexible machine learning algorithms to estimate both nuisance functions and target estimand. Theoretical properties are demonstrated when the target estimand is approximated by sieve approximation with B-splines, including error rates, asymptotic normality, and confidence intervals.

stat.ME↗

arXiv2025

The BrowserGym Ecosystem for Web Agent Research

The BrowserGym ecosystem addresses the growing need for efficient evaluation and benchmarking of web agents, particularly those leveraging automation and Large Language Models (LLMs). Many existing benchmarks suffer from fragmentation and inconsistent evaluation methodologies, making it challenging to achieve reliable comparisons and reproducible results. In an earlier work, Drouin et al. (2024) introduced BrowserGym which aims to solve this by providing a unified, gym-like environment with well-defined observation and action spaces, facilitating standardized evaluation across diverse benchmarks. We propose an extended BrowserGym-based ecosystem for web agent research, which unifies existing benchmarks from the literature and includes AgentLab, a complementary framework that aids in agent creation, testing, and analysis. Our proposed ecosystem offers flexibility for integrating new benchmarks while ensuring consistent evaluation and comprehensive experiment management. As a supporting evidence, we conduct the first large-scale, multi-benchmark web agent experiment and compare the performance of 6 state-of-the-art LLMs across 6 popular web agent benchmarks made available in BrowserGym. Among other findings, our results highlight a large discrepancy between OpenAI and Anthropic's latests models, with Claude-3.5-Sonnet leading the way on almost all benchmarks, except on vision-related tasks where GPT-4o is superior. Despite these advancements, our results emphasize that building robust and efficient web agents remains a significant challenge, due to the inherent complexity of real-world web environments and the limitations of current models.

cs.LG↗

arXiv2025

The impact of job stability on monetary poverty in Italy: causal small area estimation

Job stability - encompassing secure contracts, adequate wages, social benefits, and career opportunities - is a critical determinant in reducing monetary poverty, as it provides households with reliable income and enhances economic well-being. This study leverages EU-SILC survey and census data to estimate the causal effect of job stability on monetary poverty across Italian provinces, quantifying its influence and analyzing regional disparities. We introduce a novel causal small area estimation (CSAE) framework that integrates global and local estimation strategies for heterogeneous treatment effect estimation, effectively addressing data sparsity at the provincial level. Furthermore, we develop a general bootstrap scheme to construct reliable confidence intervals, applicable regardless of the method used for estimating nuisance parameters. Extensive simulation studies demonstrate that our proposed estimators outperform classical causal inference methods in terms of stability while maintaining computational scalability for large datasets. Applying this methodology to real-world data, we uncover significant relationships between job stability and poverty across six Italian regions, offering critical insights into regional disparities and their implications for evidence-based policy design.

stat.AP↗

arXiv2025

Simultaneous Estimation of Multiple Treatment Effects from Observational Studies

Unmeasured confounding presents a significant challenge in causal inference from observational studies. Classical approaches often rely on collecting proxy variables, such as instrumental variables. However, in applications where the effects of multiple treatments are of simultaneous interest, finding a sufficient number of proxy variables for consistent estimation of treatment effects can be challenging. Various methods in the literature exploit the structure of multiple treatments to address unmeasured confounding. In this paper, we introduce a novel approach to causal inference with multiple treatments, assuming sparsity in the causal effects. Our procedure autonomously selects treatments with non-zero causal effects, thereby providing a sparse causal estimation. Comprehensive evaluations using both simulated and Genome-Wide Association Study (GWAS) datasets demonstrate the effectiveness and robustness of our method compared to alternative approaches.

stat.ME↗

arXiv2025

Modeling Cell Developmental Trajectory using Multinomial Unbalanced Optimal Transport

Single-cell trajectory analysis aims to reconstruct the biological developmental processes of cells as they evolve over time, leveraging temporal correlations in gene expression. During cellular development, gene expression patterns typically change and vary across different cell types. A significant challenge in this analysis is that RNA sequencing destroys the cell, making it impossible to track gene expression across multiple stages for the same cell. Recent advances have introduced the use of optimal transport tools to model the trajectory of individual cells. In this paper, our focus shifts to a question of greater practical importance: we examine the differentiation of cell types over time. Specifically, we propose a novel method based on discrete unbalanced optimal transport to model the developmental trajectory of cell types. Our method detects biological changes in cell types and infers their transitions to different states by analyzing the transport matrix. We validated our method using single-cell RNA sequencing data from mouse embryonic fibroblasts. The results accurately identified major developmental changes in cell types, which were corroborated by experimental evidence. Furthermore, the inferred transition probabilities between cell types are highly congruent to biological ground truth.

stat.AP↗

arXiv2024

Functional principal component analysis with informative observation times

Functional principal component analysis has been shown to be invaluable for revealing variation modes of longitudinal outcomes, which serves as important building blocks for forecasting and model building. Decades of research have advanced methods for functional principal component analysis often assuming independence between the observation times and longitudinal outcomes. Yet such assumptions are fragile in real-world settings where observation times may be driven by outcome-related reasons. Rather than ignoring the informative observation time process, we explicitly model the observational times by a counting process dependent on time-varying prognostic factors. Identification of the mean, covariance function, and functional principal components ensues via inverse intensity weighting. We propose using weighted penalized splines for estimation and establish consistency and convergence rates for the weighted estimators. Simulation studies demonstrate that the proposed estimators are substantially more accurate than the existing ones in the presence of a correlation between the observation time process and the longitudinal outcome process. We further examine the finite-sample performance of the proposed method using the Acute Infection and Early Disease Research Program study.

stat.ME↗

arXiv2024

Stationarity of Manifold Time Series

In modern interdisciplinary research, manifold time series data have been garnering more attention. A critical question in analyzing such data is ``stationarity'', which reflects the underlying dynamic behavior and is crucial across various fields like cell biology, neuroscience and empirical finance. Yet, there has been an absence of a formal definition of stationarity that is tailored to manifold time series. This work bridges this gap by proposing the first definitions of first-order and second-order stationarity for manifold time series. Additionally, we develop novel statistical procedures to test the stationarity of manifold time series and study their asymptotic properties. Our methods account for the curved nature of manifolds, leading to a more intricate analysis than that in Euclidean space. The effectiveness of our methods is evaluated through numerical simulations and their practical merits are demonstrated through analyzing a cell-type proportion time series dataset from a paper recently published in Cell. The first-order stationarity test result aligns with the biological findings of this paper, while the second-order stationarity test provides numerical support for a critical assumption made therein.

stat.ME↗