Searcharxiv⌕ Search

arXiv subjects

Yue Liu

Publications and source records attributed to Yue Liu.

At least 73 records · Page 4Linked to original sources
arXiv2026

Enhancing Financial Report Question-Answering: A Retrieval-Augmented Generation System with Reranking Analysis

Financial analysts face significant challenges extracting information from lengthy 10-K reports, which often exceed 100 pages. This paper presents a Retrieval-Augmented Generation (RAG) system designed to answer questions about S&P 500 financial reports and evaluates the impact of neural reranking on system performance. Our pipeline employs hybrid search combining full-text and semantic retrieval, followed by an optional reranking stage using a cross-encoder model. We conduct systematic evaluation using the FinDER benchmark dataset, comprising 1,500 queries across five experimental groups. Results demonstrate that reranking significantly improves answer quality, achieving 49.0 percent correctness for scores of 8 or above compared to 33.5 percent without reranking, representing a 15.5 percentage point improvement. Additionally, the error rate for completely incorrect answers decreases from 35.3 percent to 22.5 percent. Our findings emphasize the critical role of reranking in financial RAG systems and demonstrate performance improvements over baseline methods through modern language models and refined retrieval strategies.

cs.CL↗
arXiv2026

Generalizable 3D Gaussian Splatting enabled Semantic Coding for Real-Time Immersive Video Communications

Real-time immersive video communications, particularly high-fidelity 3D telepresence, necessitates a synergistic balance between instantaneous dynamic scene reconstruction and high-efficiency data transmission. While recent advancements in feed-forward 3D Gaussian Splatting (3DGS) have enabled real-time rendering, performing multi-view video coding and 3D reconstruction in a decoupled manner leads to suboptimal compression efficiency and high computational complexity. To address this, we propose GS-SCNet, the first unified end-to-end framework that seamlessly integrates generalizable 3DGS reconstruction with a dedicated deep Semantic Coding pipeline. Our architecture is underpinned by two core technical contributions: (i) we introduce a Disparity-Guided Parallel Semantic Codec that exploits epipolar geometric priors to facilitate cross-view contextual interaction via disparity compensation and semantic fusion, thereby enabling real-time parallel processing of stereo streams while significantly enhancing rate-distortion performance, and (ii) we develop a Lightweight Gaussian Parameter Predictor which directly projects decoded semantic latents into 3DGS attributes, obviating the need for intermediate pixel-domain reconstruction. By coupling the codec with the task-specific predictor, our framework extracts geometric correlations only once, effectively eliminating the redundant computational bottleneck inherent in conventional decoupled paradigms. Extensive evaluations on both synthetic and real-world human datasets demonstrate that GS-SCNet achieves a superior trade-off across compression efficiency, rendering quality, and real-time performance. Notably, our framework exhibits strong cross-domain generalization and robustness against compression artifacts when applied to out-of-domain real-world data, significantly outperforming conventional decoupled transmission paradigms.

eess.IV↗
arXiv2026

KLong: Training LLM Agent for Extremely Long-horizon Tasks

This paper introduces KLong, an open-source LLM agent trained to solve extremely long-horizon tasks. The principle is to first cold-start the model via trajectory-splitting SFT, then scale it via progressive RL training. Specifically, we first activate basic agentic abilities of a base model with a comprehensive SFT recipe. Then, we introduce Research-Factory, an automated pipeline that generates high-quality training data by collecting research papers and constructing evaluation rubrics. Using this pipeline, we build thousands of long-horizon trajectories distilled from Claude 4.5 Sonnet (Thinking). To train with these extremely long trajectories, we propose a new trajectory-splitting SFT, which preserves early context, progressively truncates later context, and maintains overlap between sub-trajectories. In addition, to further improve long-horizon task-solving capability, we propose a novel progressive RL, which schedules training into multiple stages with progressively extended timeouts. Experiments demonstrate the superiority and generalization of KLong, as shown in Figure 1. Notably, our proposed KLong (106B) surpasses Kimi K2 Thinking (1T) by 11.28% on PaperBench, and the performance improvement generalizes to other coding benchmarks like SWE-bench Verified and MLE-bench.

cs.AI↗
arXiv2026

Scaling Coding Agents via Atomic Skills

Current LLM coding agents are predominantly trained on composite benchmarks (e.g., bug fixing), which often leads to task-specific overfitting and limited generalization. To address this, we propose a novel scaling paradigm that shifts the focus from task-level optimization to atomic skill mastery. We first formalize five fundamental atomic skills, code localization, code editing, unit-test generation, issue reproduction, and code review, that serve as the basis vectors for complex software engineering tasks. Compared with composite coding tasks, these atomic skills are more generalizable and composable. Then, we scale coding agents by performing joint RL over atomic skills. In this manner, atomic skills are consistently improved without negative interference or trade-offs between them. Notably, we observe that improvements in these atomic skills generalize well to other unseen composite coding tasks, such as bug-fixing, code refactoring, machine learning engineering, and code security. The observation motivates a new scaling paradigm for coding agents by training with atomic skills. Extensive experiments demonstrate the effectiveness of our proposed paradigm. Notably, our joint RL improves average performance by 18.7% on 5 atomic skills and 5 composite tasks.

cs.SE↗
arXiv2026

Improving the Completeness and Comparability of Segment Disclosures: A Large Language Model Approach

Segment-level disclosures are a central component of financial reporting, providing insight into firms' internal organization and the allocation of economic activities across operating units. However, segment information is often presented in both qualitative and quantitative forms, dispersed across tables and narrative sections of Form 10-K filings. Empirical research relying on structured databases faces both completeness and comparability challenges, as some firm-year observations may be missing, nested segment disclosures are not captured, and support for longitudinal and cross-firm comparability is limited. This study develops a large language model-based framework to extract segment disclosures directly from Form 10-K filings and to preserve both reportable and nested segment information. We further design a retrieval augmented system that incorporates information across multiple filings to support comparability. We use two representative settings to demonstrate its application: longitudinal analysis within a firm to interpret segment changes over time, and cross firm alignment of geographic segments across firms with different reporting structures. The results indicate that the artifact accurately extracts segment-level information and effectively addresses questions that require cross-period knowledge, demonstrating the potential of LLM-based approaches to enhance the measurement and interpretation of segment disclosures.

cs.CL↗
arXiv2026

EvoTest: Evolutionary Test-Time Learning for Self-Improving Agentic Systems

A fundamental limitation of current AI agents is their inability to learn complex skills on the fly at test time, often behaving like "clever but clueless interns" in novel environments. This severely limits their practical utility. To systematically measure and drive progress on this challenge, we first introduce the Jericho Test-Time Learning (J-TTL) benchmark. J-TTL is a new evaluation setup where an agent must play the same game for several consecutive episodes, attempting to improve its performance from one episode to the next. On J-TTL, we find that existing adaptation methods like reflection, memory, or reinforcement learning struggle. To address the challenges posed by our benchmark, we present EvoTest, an evolutionary test-time learning framework that improves an agent without any fine-tuning or gradients-by evolving the entire agentic system after every episode. EvoTest has two roles: the Actor Agent, which plays the game, and the Evolver Agent, which analyzes the episode transcript to propose a revised configuration for the next run. This configuration rewrites the prompt, updates memory by logging effective state-action choices, tunes hyperparameters, and learns the tool-use routines. On our J-TTL benchmark, EvoTest consistently increases performance, outperforming not only reflection and memory-only baselines but also more complex online fine-tuning methods. Notably, our method is the only one capable of winning two games (Detective and Library), while all baselines fail to win any.

cs.AI↗
arXiv2026

Coarse Graining Reveals a Fluctuation-theorem-like Asymmetry in Financial Markets

Fluctuation theorems show how coarse graining transforms microscopic symmetry into observable irreversibility. Here we ask whether an analogous symmetrybased diagnostic can be constructed for financial markets. At the microscopic level, each transaction pairs a buyer and a seller, whereas trading decisions are typically made from coarse-grained price histories. Using symmetric takeprofit and stop-loss rules, we compare the holding-time distributions of long and short trading ensembles generated from the same price series. Across equityindices, individual stocks and cryptocurrencies, the log-ratio of the two distributions shows a robust crossover. It remains nearly constant at short durations but becomes linear in the tail, implying an exponential directional asymmetry. The tail slope defines an effective market temperature, an operational measure of fluctuation intensity on the chosen observation scale. A Bachelier first-passage benchmark captures the exponential tails but not the asymmetry, because long and short positions share the same leading decay rate. By contrast, short-time correlations between overlapping positions provide a minimal mechanism for the asymmetry by generating direction-dependent subleading relaxation spectra in a coarse-grained Markov description. Together, these results establish a fluctuation-theorem-like diagnostic of irreversibility in financial markets and, more broadly, in complex systems accessible only through coarse-grained observables.

cond-mat.stat-mech↗
arXiv2026

RadarSplat-RIO: Indoor Radar-Inertial Odometry with Gaussian Splatting-Based Radar Bundle Adjustment

Radar is more resilient to adverse weather and lighting conditions than visual and Lidar simultaneous localization and mapping (SLAM). However, most radar SLAM pipelines still rely heavily on frame-to-frame odometry, which leads to substantial drift. While loop closure can correct long-term errors, it requires revisiting places and relies on robust place recognition. In contrast, visual odometry methods typically leverage bundle adjustment (BA) to jointly optimize poses and map within a local window. However, an equivalent BA formulation for radar has remained largely unexplored. We present the first radar BA framework enabled by Gaussian Splatting (GS), a dense and differentiable scene representation. Our method jointly optimizes radar sensor poses and scene geometry using full range-azimuth-Doppler data, bringing the benefits of multi-frame BA to radar for the first time. When integrated with an existing radar-inertial odometry frontend, our approach significantly reduces pose drift and improves robustness. Across multiple indoor scenes, our radar BA achieves substantial gains over the prior radar-inertial odometry, reducing average absolute translational and rotational errors by 90% and 80%, respectively.

cs.RO↗
arXiv2026

WebAgentGuard: A Reasoning-Driven Guard Model for Detecting Prompt Injection Attacks in Web Agents

Web agents powered by vision-language models (VLMs) enable autonomous interaction with web environments by perceiving and acting on both visual and textual webpage content to accomplish user-specified tasks. However, they are highly vulnerable to prompt injection attacks, where adversarial instructions embedded in HTML or rendered screenshots can manipulate agent behavior and lead to harmful outcomes such as information leakage. Existing defenses, including system prompt defenses and direct fine-tuning of agents, have shown limited effectiveness. To address this issue, we propose a defense framework in which a web agent operates in parallel with a dedicated guard agent, decoupling prompt injection detection from the agent's own reasoning. Building on this framework, we introduce WebAgentGuard, a reasoning-driven, multimodal guard model for prompt injection detection. We construct a synthetic multimodal dataset using GPT-5 spanning 164 topics and 230 visual and UI design styles, and train the model via reasoning-intensive supervised fine-tuning followed by reinforcement learning. Experiments across multiple benchmarks show that WebAgentGuard consistently outperforms strong baselines while preserving agent utility, without introducing additional latency.

cs.CR↗
arXiv2026

Quantifying Individual Risk for Binary Outcomes

Understanding treatment effect heterogeneity is crucial for reliable decision-making in treatment evaluation and selection. The conditional average treatment effect (CATE) is widely used to capture treatment effect heterogeneity induced by observed covariates and to design individualized treatment policies. However, it is an average metric within subpopulations, which prevents it from revealing individual risk, potentially leading to misleading results. This article fills this gap by examining individual risk for binary outcomes, specifically focusing on the fraction negatively affected (FNA), a metric that quantifies the percentage of individuals experiencing worse outcomes under treatment compared with control. Even under the strong ignorability assumption, FNA is still unidentifiable, and the existing Fréchet--Hoeffding bounds are often too wide and attainable only under extreme data-generating processes. By invoking mild conditions on the value range of the Pearson correlation coefficient between potential outcomes, we obtain improved bounds compared with the Fréchet--Hoeffding bounds. We show that paradoxically, even with a positive CATE, the lower bound on FNA can be positive, i.e., in the best-case scenario, many individuals will be harmed if they receive treatment. Additionally, we establish a nonparametric sensitivity analysis framework for FNA using the Pearson correlation coefficient as the sensitivity parameter. Furthermore, we propose nonparametric estimators for the refined FNA bounds and prove their consistency and asymptotic normality. We use simulation to evaluate the performance of the proposed estimators and apply the method to a canonical observational study.

stat.ME↗
arXiv2026

Evaluating Surrogates in Individualized Treatment Rules

In many decision-making problems, the primary outcome is expensive, time-consuming, or difficult to observe, so individualized treatment rules (ITRs) may be instead learned from surrogate endpoints. However, a surrogate that is highly associated with the primary outcome, or even satisfies existing surrogate criteria, may not necessarily induce a treatment rule that performs well on the primary outcome, especially under treatment resource budget constraints. In this paper, we develop a principled framework for evaluating the decision-making value of surrogate endpoints. We introduce three ITR-oriented performance measures: surrogate regret, which assesses the expected loss from using the surrogate-optimal ITR instead of outcome-optimal ITR; surrogate gain, which quantifies the benefit of surrogate-optimal ITRs relative to the no-treatment baseline; and surrogate efficiency, which evaluates improvement over random treatment assignment. We also extend them to budget-constrained settings. We propose augmented inverse probability weighted (AIPW) estimators for these measures and establish their large-sample properties. We demonstrate the proposed approach on both simulations and an application to the Criteo dataset.

stat.ME↗
arXiv2026

Robustness via Referencing: Defending against Prompt Injection Attacks by Referencing the Executed Instruction

Large language models (LLMs) have demonstrated impressive performance and have come to dominate the field of natural language processing (NLP) across various tasks. However, due to their strong instruction-following capabilities and inability to distinguish between instructions and data content, LLMs are vulnerable to prompt injection attacks. These attacks manipulate LLMs into deviating from the original input instructions and executing maliciously injected instructions within data content, such as web documents retrieved from search engines. Existing defense methods, including prompt-engineering and fine-tuning approaches, typically instruct models to follow the original input instructions while suppressing their tendencies to execute injected instructions. However, our experiments reveal that suppressing instruction-following tendencies is challenging. Through analyzing failure cases, we observe that although LLMs tend to respond to any recognized instructions, they are aware of which specific instructions they are executing and can correctly reference them within the original prompt. Motivated by these findings, we propose a novel defense method that leverages, rather than suppresses, the instruction-following abilities of LLMs. Our approach prompts LLMs to generate responses that include both answers and their corresponding instruction references. Based on these references, we filter out answers not associated with the original input instructions. Comprehensive experiments demonstrate that our method outperforms prompt-engineering baselines and achieves performance comparable to fine-tuning methods, reducing the attack success rate (ASR) to 0 percent in some scenarios. Moreover, our approach has minimal impact on overall utility.

cs.CR↗
arXiv2026

DiffuGuard: How Intrinsic Safety is Lost and Found in Diffusion Large Language Models

The rapid advancement of Diffusion Large Language Models (dLLMs) introduces unprecedented vulnerabilities that are fundamentally distinct from Autoregressive LLMs, stemming from their iterative and parallel generation mechanisms. In this paper, we conduct an in-depth analysis of dLLM vulnerabilities to jailbreak attacks across two distinct dimensions: intra-step and inter-step dynamics. Experimental results reveal a harmful bias inherent in the standard greedy remasking strategy and identify a critical phenomenon we term Denoising-path Dependence, where the safety of early-stage tokens decisively influences the final output. These findings also indicate that while current decoding strategies constitute a significant vulnerability, dLLMs possess a substantial intrinsic safety potential. To unlock this potential, we propose DiffuGuard, a training-free defense framework that addresses vulnerabilities through a dual-stage approach: Stochastic Annealing Remasking dynamically introduces controlled randomness to mitigate greedy selection bias, while Block-level Audit and Repair exploits internal model representations for autonomous risk detection and guided correction. Comprehensive experiments on four dLLMs demonstrate DiffuGuard's exceptional effectiveness, reducing Attack Success Rate against six diverse jailbreak methods from 47.9% to 14.7% while preserving model utility and efficiency. Our code is available at: https://github.com/niez233/DiffuGuard.

cs.CL↗
arXiv2026

Beyond the Academic Monoculture: A Unified Framework and Industrial Perspective for Attributed Graph Clustering

Attributed Graph Clustering (AGC) is a fundamental unsupervised task that partitions nodes into cohesive groups by jointly modeling structural topology and node attributes. While the advent of graph neural networks and self-supervised learning has catalyzed a proliferation of AGC methodologies, a widening chasm persists between academic benchmark performance and the stringent demands of real-world industrial deployment. To bridge this gap, this survey provides a comprehensive, industrially grounded review of AGC from three complementary perspectives. First, we introduce the Encode-Cluster-Optimize taxonomic framework, which decomposes the diverse algorithmic landscape into three orthogonal, composable modules: representation encoding, cluster projection, and optimization strategy. This unified paradigm enables principled architectural comparisons and inspires novel methodological combinations. Second, we critically examine prevailing evaluation protocols to expose the field's academic monoculture: a pervasive over-reliance on small, homophilous citation networks, the inadequacy of supervised-only metrics for an inherently unsupervised task, and the chronic neglect of computational scalability. In response, we advocate for a holistic evaluation standard that integrates supervised semantic alignment, unsupervised structural integrity, and rigorous efficiency profiling. Third, we explicitly confront the practical realities of industrial deployment. By analyzing operational constraints such as massive scale, severe heterophily, and tabular feature noise alongside extensive empirical evidence from our companion benchmark, we outline actionable engineering strategies. Furthermore, we chart a clear roadmap for future research, prioritizing heterophily-robust encoders, scalable joint optimization, and unsupervised model selection criteria to meet production-grade requirements.

cs.LG↗
arXiv2026

Aligning Paralinguistic Understanding and Generation in Speech LLMs via Multi-Task Reinforcement Learning

Speech large language models (LLMs) observe paralinguistic cues such as prosody, emotion, and non-verbal sounds--crucial for intent understanding. However, leveraging these cues faces challenges: limited training data, annotation difficulty, and models exploiting lexical shortcuts over paralinguistic signals. We propose multi-task reinforcement learning (RL) with chain-of-thought prompting that elicits explicit affective reasoning. To address data scarcity, we introduce a paralinguistics-aware speech LLM (PALLM) that jointly optimizes sentiment classification from audio and paralinguistics-aware response generation via a two-stage pipeline. Experiments demonstrate that our approach improves paralinguistics understanding over both supervised baselines and strong proprietary models (Gemini-2.5-Pro, GPT-4o-audio) by 8-12% on Expresso, IEMOCAP, and RAVDESS. The results show that modeling paralinguistic reasoning with multi-task RL is crucial for building emotionally intelligent speech LLMs.

cs.CL↗
arXiv2026

AudioTrust: Benchmarking the Multifaceted Trustworthiness of Audio Large Language Models

The rapid development and widespread adoption of Audio Large Language Models (ALLMs) demand rigorous evaluation of their trustworthiness. However, existing evaluation frameworks are primarily designed for text and fail to capture vulnerabilities introduced by the acoustic properties of audio. We find that significant trustworthiness risks in ALLMs arise from non-semantic acoustic cues, such as timbre, accent, and background noise, which can be exploited to manipulate model behavior. To address this gap, we propose AudioTrust, the first large-scale and systematic framework for evaluating ALLM trustworthiness under audio-specific risks. AudioTrust covers six key dimensions: fairness, hallucination, safety, privacy, robustness, and authenticition. It includes 26 sub-tasks and a curated dataset of more than 4,420 audio samples collected from real-world scenarios, including daily conversations, emergency calls, and voice assistant interactions, and is specifically designed to probe trustworthiness across multiple dimensions. Our comprehensive evaluation spans 18 experimental settings and uses human-validated automated pipelines to enable objective and scalable assessment of model outputs. Experimental results on 14 state-of-the-art open-source and closed-source ALLMs reveal important limitations and failure boundaries under diverse high-risk audio scenarios, providing critical insights for the secure and trustworthy deployment of future audio models. Our platform and benchmark are publicly available at https://github.com/JusperLee/AudioTrust.

cs.SD↗
arXiv2026

VPI-Bench: Visual Prompt Injection Attacks for Computer-Use Agents

Computer-Use Agents (CUAs) with full system access enable powerful task automation but pose significant security and privacy risks due to their ability to manipulate files, access user data, and execute arbitrary commands. While prior work has focused on browser-based agents and HTML-level attacks, the vulnerabilities of CUAs remain underexplored. In this paper, we investigate Visual Prompt Injection (VPI) attacks, where malicious instructions are visually embedded within rendered user interfaces, and examine their impact on both CUAs and Browser-Use Agents (BUAs). We propose VPI-Bench, a benchmark of 306 test cases across five widely used platforms, to evaluate agent robustness under VPI threats. Each test case is a variant of a web platform, designed to be interactive, deployed in a realistic environment, and containing a visually embedded malicious prompt. Our empirical study shows that current CUAs and BUAs can be deceived at rates of up to 51% and 100%, respectively, on certain platforms. The experimental results also indicate that system prompt defenses offer only limited improvements. These findings highlight the need for robust, context-aware defenses to ensure the safe deployment of multimodal AI agents in real-world environments. The code and dataset are available at: https://github.com/cua-framework/agents

cs.AI↗
arXiv2026

IMMACULATE: A Practical LLM Auditing Framework via Verifiable Computation

Commercial large language models are typically deployed as black-box API services, requiring users to trust providers to execute inference correctly and report token usage honestly. We present IMMACULATE, a practical auditing framework that detects economically motivated deviations-such as model substitution, quantization abuse, and token overbilling-without trusted hardware or access to model internals. IMMACULATE selectively audits a small fraction of requests using verifiable computation, achieving strong detection guarantees while amortizing cryptographic overhead. Experiments on dense and MoE models show that IMMACULATE reliably distinguishes benign and malicious executions with under 1% throughput overhead. Our code is published at https://github.com/guo-yanpei/Immaculate.

cs.CR↗