arXiv subjects

Yujing Lu

Publications and source records attributed to Yujing Lu.

5 recordsLinked to original sources

arXiv2026

TRE: Encouraging Exploration in the Trust Region

Entropy regularization is a standard technique in reinforcement learning (RL) to enhance exploration, yet it yields negligible effects or even degrades performance in Large Language Models (LLMs). We attribute this failure to the cumulative tail risk inherent to LLMs with massive vocabularies and long generation horizons. In such environments, standard global entropy maximization indiscriminately dilutes probability mass into the vast tail of invalid tokens rather than focusing on plausible candidates, thereby disrupting coherent reasoning. To address this, we propose Trust Region Entropy (TRE), a method that encourages exploration strictly within the model's trust region. Extensive experiments across mathematical reasoning (MATH), combinatorial search (Countdown), and preference alignment (HH) tasks demonstrate that TRE consistently outperforms vanilla PPO, standard entropy regularization, and other exploration baselines. Our code is available at https://github.com/WhyChaos/TRE-Encouraging-Exploration-in-the-Trust-Region.

cs.CL↗

arXiv2025

Penalized mixed models to adjust for batch effects and unobserved confounding in high dimensional regression

Confounding can lead to spurious associations. Typically, one must observe confounders in order to adjust for them, but in high-dimensional settings, recent research has shown that it becomes possible to adjust even for unobserved confounders. The methods for carrying out these adjustments, however, have not been thoroughly investigated. In this study, we derive a confounding framework in which the signal, bias, and variability can be cleanly partitioned and quantified, thereby enabling simulations in which one varies the bias-to-signal ratio while holding the signal-to-noise ratio fixed. Using this construction, we demonstrate the impact of the amount and complexity of unobserved confounding on the performance of competing methods, including the LASSO, principal components LASSO (PC-LASSO), and penalized linear mixed models (PLMMs). We identify scenarios in which each method outperforms the others and find that overall, PLMM is the most robust approach.

stat.ME↗

arXiv2025

How Auxiliary Reasoning Unleashes GUI Grounding in VLMs

Graphical user interface (GUI) grounding is a fundamental task for building GUI agents. However, general vision-language models (VLMs) struggle with this task due to a lack of specific optimization. We identify a key gap in this paper: while VLMs exhibit significant latent grounding potential, as demonstrated by their performance measured by Pointing Game, they underperform when tasked with outputting explicit coordinates. To address this discrepancy and bypass the high data and annotation costs of current fine-tuning approaches, we propose three zero-shot auxiliary reasoning methods. By providing explicit spatial cues such as axes, grids and labeled intersections as part of the input image, these methods enable VLMs to better articulate their implicit spatial understanding capabilities. We evaluate these methods on four GUI grounding benchmarks across seven open-source and proprietary VLMs. Experimental results show substantial gains from auxiliary reasoning. Mark-Grid Scaffold boosts Gemini-3.1-Pro from 11.72\% under direct inference to 95.20\% on ScreenSpot-v2, achieves state-of-the-art performance on ScreenSpot, and approaches the strongest fine-tuned methods on ScreenSpot-v2 and UI-I2E-Bench. Our code is available at https://github.com/liweim/AuxiliaryReasoning.

cs.CV↗

arXiv2025

DomainCQA: Crafting Knowledge-Intensive QA from Domain-Specific Charts

Chart Question Answering (CQA) evaluates Multimodal Large Language Models (MLLMs) on visual understanding and reasoning over chart data. However, existing benchmarks mostly test surface-level parsing, such as reading labels and legends, while overlooking deeper scientific reasoning. We propose DomainCQA, a framework for constructing domain-specific CQA benchmarks that emphasize both visual comprehension and knowledge-intensive reasoning. It integrates complexity-aware chart selection, multitier QA generation, and expert validation. Applied to astronomy, DomainCQA yields AstroChart, a benchmark of 1,690 QA pairs over 482 charts, exposing persistent weaknesses in fine-grained perception, numerical reasoning, and domain knowledge integration across 21 MLLMs. Fine-tuning on AstroChart improves performance across fundamental and advanced tasks. Pilot QA sets in biochemistry, economics, medicine, and social science further demonstrate DomainCQA's generality. Together, our results establish DomainCQA as a unified pipeline for constructing and augmenting domain-specific chart reasoning benchmarks.

cs.CL↗

arXiv2025

plmmr: an R package to fit penalized linear mixed models for genome-wide association data with complex correlation structure

Correlation among the observations in high-dimensional regression modeling can be a major source of confounding. We present a new open-source package, plmmr, to implement penalized linear mixed models in R. This R package estimates correlation among observations in high-dimensional data and uses those estimates to improve prediction with the best linear unbiased predictor. The package uses memory-mapping so that genome-scale data can be analyzed on ordinary machines even if the size of data exceeds RAM. We present here the methods, workflow, and file-backing approach upon which plmmr is built, and we demonstrate its computational capabilities with two examples from real GWAS data.

stat.CO↗