论文检索

Social Aspects · Safety

Khang Tran, Yazan Boshmaf, Issa Khalil, Hai Phan, Ting Yu, Md Rizwan Parvez

Code Large Language Models (CLLMs) serve as the core of modern code agents, enabling developers to automate complex software development tasks. In this paper, we present Poison-with-Style (PwS), a practical and stealthy model poisoning attack targeting CLLMs. Unlike prior attacks that assume an active adversary capable of directly embedding explicit triggers (e.g., specific words) into developers' prompts during inference, PwS leverages developers' code styles as covert triggers implicitly embedded within their prompts. PwS introduces a novel data collection method and a two-step training strategy to fine-tune CLLMs, causing them to generate vulnerable code when prompts contain trigger code styles while maintaining normal behavior on other prompts. Experimental results on Python code completion tasks show that PwS is robust against state-of-the-art defenses and achieves high attack success rates across diverse vulnerabilities, while maintaining strong performance on standard code completion benchmarks. For example, in code completion tasks that are vulnerable to improper input validation (i.e., CWE-20), the poisoned model generates insecure code up to 95\% of the cases when the trigger code style is used, with less than $5\%$ drop in pass@1 performance on the HumanEval and MBPP benchmarks.

Social Aspects · Safety

Seokil Ham, Jaehyuk Jang, Wonjun Lee, Changick Kim

Fine-tuning-as-a-Service (FaaS) enables personalization of large language models (LLMs) but poses significant safety risks, as fine-tuning user-provided data degrades the model's safety-alignment. Prior works addressing this issue typically rely on explicit regularization, which leads to practical limitations. In this paper, we propose a different paradigm that neutralizes harmful updates via harmful gradient saturation rather than explicit suppression. Our key observation is that, in a jailbroken LLM, safety-degrading gradients are largely saturated, while gradients unrelated to safety remain active during fine-tuning. Based on this insight, we introduce a **BufferLoRA-based fine-tuning framework**. BufferLoRA is a removable adapter that temporarily jailbreaks the model during user fine-tuning, saturating harmful updates while allowing a UserLoRA to learn user-specific tasks. After fine-tuning, BufferLoRA is removed to restore the base model’s original safety-alignment. To further reinforce safety, we additionally train a SafetyLoRA and integrate its safety components into the UserLoRA via QR decomposition-based merging strategy. Extensive experiments show that our framework achieves superior performance in both safety and utility, without requiring additional safety data during fine-tuning and with minimal computational cost.

Deep Learning · Other Representation Learning

Sigurd Gaukstad, Erik Hermansen, Valdemar Kargård Olsen, Melvin Vaupel, Benjamin Dunn

Deep neural networks learn representations where individual features often lack interpretable meaning; a single neuron may activate for scattered, unrelated inputs. We introduce coherence, a geometric property inspired by neural coding in the brain, where neurons like grid cells and head direction cells respond to contiguous regions of state space. A non-negative matrix is coherent if both each row (sample) attends to geometrically clustered columns (features) and, vice versa, each feature attends to clustered samples. We prove that coherent matrices induce a bounded interleaving between the Vietoris-Rips filtrations of samples and features, guaranteeing that both spaces share compatible topological structure. This geometric constraint facilitates interpretability. For example, if data lies on a circle, coherent features must tile that circle into contiguous arcs. We introduce COH, a differentiable regularizer based on Fréchet variance that enforces coherence during training. Unlike sparsity, which bounds how many samples a feature activates on, coherence bounds *which* samples, requiring geometric connectivity rather than only rarity. This yields not just interpretable features but an interpretable feature space. We validate COH using synthetic and rotated MNIST datasets.

Reinforcement Learning · Planning

Jiangweizhi Peng, Yuanxin Liu, Ruida Zhou, Charles Fleming, Zhaoran Wang, Alfredo Garcia, Mingyi Hong

Training LLMs as interactive agents for multi-turn decision-making remains challenging, particularly in long-horizon tasks with sparse and delayed rewards, where agents must execute extended sequences of actions before receiving meaningful feedback. Most existing reinforcement learning (RL) methods model LLM agents as *flat* policies operating at a single time scale, selecting an action at each turn. In sparse-reward settings, this forces the agent to infer long-range dependencies solely from distant end-of-trajectory signals, often leading to inefficient learning and unstable behavior in complex environments. We propose **HiPER**, a novel **Hi**erarchical **P**lan–**E**xecute **R**L framework that jointly models and optimizes high-level subgoal planning and low-level action execution for LLM agents to overcome flat RL's brittle long-horizon behavior and weak credit assignment under sparse outcome feedback. By maintaining persistent subgoals across multiple turns and explicitly deciding when to switch between them, HiPER introduces structured intermediate decision points that facilitate learning under sparse feedback, converting implicit multi-turn structure into learnable decisions at different time scales. To enable effective training, we introduce Hierarchical Advantage Estimation (HAE), a two-timescale policy gradient method that assigns credit to both action execution and subgoal transitions and achieves variance reduction relative to flat advantage estimation. Empirically, HiPER achieves state-of-the-art performance on challenging interactive benchmarks, reaching 97.4\% success on ALFWorld (+6.6\% over the best prior method) and 83.3\% on WebShop, with especially large gains on long-horizon tasks requiring multiple dependent subtasks. These results highlight the importance of explicit hierarchical decomposition for scalable RL training of multi-turn LLM agents.

Social Aspects · Alignment

Matthew Bozoukov, Matthew Nguyen, Dani Roytburg, Jou Barzdukas, Mackenzie Puig-Hall, Narmeen Oozeer

Recent research has shown that large language models (LLM) favor own outputs when acting as judges, undermining the integrity of automated post-training and evaluation workflows. However, it is difficult to disentangle which evaluation biases are explained by narcissism versus general experimental confounds, distorting measurements of self-preference bias. We discover a core methodological confound which could reduce measurement error by **89.6%**. Specifically, LLM evaluators may deliver self-preferring verdicts when the judge responds to queries which they completed incorrectly themselves; this would be true regardless of whether one of their responses is their own. To decouple self-preference signals from noisy outputs on hard problems, we introduce an **Evaluator Quality Baseline**, which compares the probability that a judge incorrectly votes for itself against the probability that it votes for an incorrect response from another model. Evaluating this simple baseline on 37,448 queries, only **51%** of initial findings retain statistical significance. Finally, we turn towards characterizing the entropy of "easy" versus "hard" evaluation votes from LLM judges. Our corrective baseline enables future research on self-preference by eliminating noisy data from potential solutions. More widely, this work contributes to the growing body of work on cataloging and isolating judge-bias effects

Applications · Computer Vision

Jundong Zhou, Tianao Cai, Yujie Huang, Xinbing Wang, Guang-Zhong Yang, Nanyang Ye

Large Vision–Language Models (LVLMs) have demonstrated remarkable capabilities across diverse multimodal tasks, yet their high inference costs necessitate low-bit deployment. Existing post-training quantization (PTQ) pipelines primarily adopt methodologies from text-only LLMs by treating multimodal inputs as homogeneous sequences, overlooking the heterogeneous information density inherent in LVLMs. In this work, we present ImpQuant, an importance-aware PTQ framework tailored for LVLMs that mitigates low-bit accuracy degradation via fine-grained token-importance reweighted calibration and outlier-aware activation quantization. Our key insight is that quantization errors on decision-critical tokens disproportionately impact overall model behavior. Accordingly, we reweight the calibration loss using aggregated attention for textual tokens and a contextual redundancy metric for visual tokens, respectively. Across multiple LVLM backbones and diverse multimodal benchmarks, our approach consistently improves accuracy at low bitwidth and reduces quantization-induced object hallucinations compared to state-of-the-art PTQ baselines.

Deep Learning · Large Language Models

Shankar Padmanabhan, Mustafa Omer Gul, Tanya Goyal

Post-training endows pretrained LLMs with a variety of desirable skills, such as instruction-following, reasoning, and others. However, these post-trained LLMs only encode knowledge up to a cut-off date, necessitating continual adaptation. Unfortunately, existing solutions cannot effectively learn new knowledge from adaptation document corpora and simultaneously mitigate the forgetting of earlier learned capabilities. To address this, we introduce Distillation via Split Contexts (DiSC), a simple context-distillation based approach for continual knowledge adaptation. DiSC derives student and teacher distributions by conditioning on distinct segments of the training example and minimizes the KL divergence between them for the common tokens. This insight allows us to efficiently apply context-distillation without requiring explicit generation steps during training. We run experiments on three post-trained models and two adaptation domains. Compared to prior finetuning and distillation methods for continual adaptation, DiSC consistently reports the best trade-off between learning new knowledge and mitigating forgetting of previously learned skills like instruction-following and reasoning, or factual knowledge.

Social Aspects · Accountability, Transparency, and Interpretability

Zehao Wang, shilong jin, Zhao Cao, Lanjun Wang

LLM-based multi-agent systems can fail even when planned actions are executed correctly because agents may misjudge their knowledge when evaluating plan feasibility, a phenomenon we term epistemic miscalibration in planning. Unlike execution errors, epistemic miscalibration is latent during planning, as generated plans can remain self-consistent and executable without observable errors; the miscalibration is also dynamic, as new information can alter feasibility assessments, potentially obscuring past miscalibration signals and causing them to recur over time. To address this, we propose the Epistemic Planning Calibration Agentic Workflow (EPC-AW), which assesses whether plans remain supported under varying information conditions rather than directly verifying feasibility. EPC-AW employs Information-consistency-based Plan Selection, selecting plans whose evaluations are stable across agents, together with Consistency-guided Epistemic State Refinement to adapt calibration over time by leveraging past discrepancies to guide future planning. Experiments show that EPC-AW improves system-level success by an average of 9.75\%.

Deep Learning · Other Representation Learning

Manuel Hirth, Lukas Koberg, Nasser Jazdi, Enkelejda Kasneci

While diffusion models have advanced anomaly detection, their reliance on multi-step noise schedules introduces significant computational complexity. In this paper, we demonstrate that the generative capability of diffusion is not required for tabular one-class anomaly detection. We revisit core principles of denoising without targeting data generation and present a deep-learning approach that streamlines these objectives into a fixed-noise formulation. Unlike standard denoising autoencoders that rely on reconstruction error, our method utilizes a preconditioning with an explicit linear reference channel. We train a denoising predictor to recover clean samples from perturbed observations and derive anomaly scores from the expected deviation under repeated perturbations. We theoretically motivate this with a stability proxy by a first-order approximation rather than just distance to the manifold. On the well-established ADBench benchmark, our approach achieves improvements over existing methods of 1.22\% in AUCROC and 1.13\% in AUCPR, the most informative and threshold-independent metrics. Our approach emphasizes structural simplicity and efficiency, indicating that a single-step, stability-based objective outperforms complex generative schedules for tabular data.

Jun-Peng Jiang, Shiyin Lu, An-Yang Ji, Yinglun Li, Qing-Guo Chen, Zhao Xu, Weihua Luo, Kaifu Zhang, De-Chuan Zhan, Han-Jia Ye

This paper presents OvisOCR, a lightweight and strictly end-to-end Multimodal Language Model (MLLM) tailored for document parsing. Unlike current methods that rely on complex "Crop-OCR-Merge" cascades to handle high-resolution inputs, OvisOCR directly maps full-page visual signals to structured Markdown without localized slicing or layout detection dependencies. Through extensive evaluations on the OmniDoc benchmark, OvisOCR achieves SOTA performance, demonstrating that a compact E2E model can effectively ``digest'' the capabilities of intricate pipelines and surpass specialized and general methods. Technically, OvisOCR establishes a holistic paradigm that synergizes specialized perception with general reasoning, distilling fine-grained recognition from OCR engines and semantic correction from LLMs into a unified model. To balance the performance across diverse document constituents, we design category-specific reward mechanisms for distinct element types, such as dense text, complex tables, and formulas, and ensure the model enhances its formatting strengths for each modality concurrently. This approach effectively resolves the optimization conflict, guaranteeing that improvements in structural layout parsing do not come at the expense of omitting fine-grained textual details. Empirical results confirm that OvisOCR eliminates the error propagation inherent in split-and-merge architectures, offering a streamlined path for next-generation document intelligence.

Theory · Deep Learning

Bodie Rubacher

Softmax attention is increasingly treated as a reusable inference primitive in transformers. Much prior theory covers Gaussian/linear models or assumes one shared quadratic query--key metric, which can fail for non-Gaussian exponential-family likelihoods with varying curvature. This risks overreading learned similarities as a global metric or as proof of Bayes-optimality. For discrete latent symbols with exponential-family observations, we show a single head can implement Bayes posteriors (and posterior means) by setting logits to log prior plus log likelihood, and we characterize single-head posteriors as exactly the log-linear (exponential-family) class. Using convex duality, we rewrite log-likelihoods as negative Bregman divergences on mean/sufficient-statistic space, making Bayes rule a soft nearest-neighbor computation; this yields a sharp boundary for globally shared quadratic metrics and a multi-head curvature-atlas approximation with head-count scaling, and we extend these guarantees to in-context estimation with consistency and finite-sample stability bounds. In synthetic Gaussian and Bernoulli in-context estimation, trained minimal attention models validate these predictions: performance approaches a Bayes-oracle baseline as prompt length grows; learned metrics align with noise precision in the Gaussian case, while Bernoulli retains a gap consistent with curvature variation. Together, these results explain when Fisher geometry should emerge, when a single metric is justified, and when multiple heads are necessary for Bayes-like in-context estimators beyond Gaussians.

Applications · Everything Else

shilong jin, Lanjun Wang, Zhuosheng Zhang

Autonomous Graphical User Interface (GUI) agents often struggle with multi-step tasks due to constrained context windows and static policies that fail to adapt to dynamic environments. To address these limitations, this work proposes the Self-Evolving GUI Agent (SE-GA), a novel framework that integrates hierarchical memory structures with an iterative self-improvement mechanism. At the core of our approach is Test-Time Memory Extension (TTME), which facilitates long-term planning by dynamically retrieving episodic, semantic, and experiential memories to provide salient contexts during inference. To ensure continuous learning, we introduce Memory-Augmented Self-Evolution (MASE), which is a training pipeline that adopts the data collected by TTME to stabilize and enhance the agent's foundational policy. Extensive evaluations across both offline and online benchmarks demonstrate SE-GA achieves state-of-the-art performance, reaching success rates of 89.0\% on ScreenSpot and 75.8\% on the challenging AndroidControl-High dataset. Furthermore, significant improvements on the AndroidWorld benchmark highlight the superior generalization to dynamic environments.

General Machine Learning · Representation Learning

Rong Hu, Ling Chen

Disentangled representation learning is a powerful paradigm for robust attribute prediction. While recent methods address attribute correlations, hidden correlations remain underexplored, where data under the value of a certain attribute exhibit underlying modes correlated with other attributes. To preserve mode information and achieve disentanglement, we jointly discover modes and enforce mode-based conditional independence. Yet, the interdependency between these two modules may lead to error amplification under naive iterations. We propose Coordinated Disentanglement with Iterative mode Discovery (CoDID), an end-to-end framework featuring a dynamic architecture that adapts to evolving number of modes, and a coordination mechanism that mitigates error amplification via meta-optimization. Empirical results demonstrate the state-of-the-art performance on diverse tasks. Codes are available at anonymous Github https://anonymous.4open.science/r/CoDID-B038.

Social Aspects · Accountability, Transparency, and Interpretability

Soyeon Kim, Seongwoo Lim, Kyowoon Lee, Jaesik Choi

Feature attribution is central to diagnosing and trusting deep neural networks, and Integrated Gradients (IG) is widely used due to its axiomatic properties. However, IG can yield unreliable explanations when the integration path between a baseline and the input passes through regions with noisy gradients. While Guided Integrated Gradients reduces this sensitivity by adaptively updating low-gradient-magnitude features, input-space guidance still produces intermediate inputs that deviate from the data manifold. To address this limitation, we propose **Manifold-Aligned Guided Integrated Gradients** (MA-GIG), which constructs attribution paths in the latent space of a pre-trained variational autoencoder. By ensuring that decoded intermediate images remain aligned with the data manifold, MA-GIG constrains gradient evaluation to statistically valid regions. Through qualitative and quantitative evaluations, we demonstrate that MA-GIG produces faithful explanations by aggregating gradients on path features proximal to the input. Consequently, our method suppresses off-manifold noise and outperforms prior path-based attribution methods across multiple datasets and classifiers.

Deep Learning · Large Language Models

Haoren Zhao, Tianyi Chen, Zhen Wang

Multimodal Large Language Models (MLLMs) have revolutionized GUI automation, yet their efficacy is largely established on idealized, single-layer interfaces. This paper identifies a critical reliability gap: state-of-the-art agents face distinct robustness challenges in real-world desktop environments characterized by multi-window stacking, occlusion, and visual clutter. To address this, we introduce WinDeskGround, a novel benchmark and synthesis framework tailored for evaluating GUI grounding robustness. Unlike static datasets, our framework parametrically generates complex desktop scenarios by controlling window occlusion, layout density, and semantic similarity, thereby simulating the distribution shifts of authentic workflows. We construct a diverse meta-dataset of 1,356 high-fidelity instruction-target pairs and conduct comprehensive evaluations of five leading MLLMs. Our results demonstrate that while top-tier agents excel in simplified settings, their accuracy declines under partial occlusion. WinDeskGround provides a valuable benchmark to facilitate the assessment and advancement of GUI agent robustness in realistic environments.

Social Aspects · Robustness

Soroush H. Zargarbashi, Mohammad Sadegh Akhondzadeh, Aleksandar Bojchevski

Robust conformal prediction (RCP) addresses confidence miscalibration in machine learning models by producing prediction sets with guaranteed coverage — these sets are guaranteed to include the true label with a user-specified high probability, even under worst-case noise. Recent works use randomized smoothing, as it provides robustness for black-box models at larger radii. Currently, there exist two setups for smoothing-based RCP: one requires extensive Monte Carlo sampling at calibration and test time but results in smaller prediction sets; the other setup produces larger prediction sets but uses a single sample at both stages. Since calibration is a one-time preprocessing step, it can accommodate substantially higher computational overhead than inference. Inspired by that, we offer procedures in between: we increase the sample rate at calibration time while using only one or few samples at test time. Increased calibration-time sampling can reduce the size of the prediction sets. With a large enough test set (which is often the case in production), our Front-Loaded RCPs have the same computational complexity as the state of the art, while producing considerably smaller sets at larger radii.

Applications · Computer Vision

Renjie Ding, Yaonan Wang, Min Liu, Jialin Zhu, Jiazheng Wang, Jiahao Zhao, tan xiao, Feixiang He, xiang chen

3D Gaussian Splatting (3DGS) enables real-time, high-fidelity view synthesis via explicit scene representations and has recently been extended to dynamic scene modeling. In spite of excellent quality and interpretability, we find explicit Dynamic GS often exhibits generalization degradation in large-motion scenes. Motivated by generalization insights from deep learning and the characteristics of Gaussian primitive optimization, we propose an update mixture strategy. This work focuses on two representative open-source explicit Dynamic GS pipelines and our approach includes: (i) a space–time dependent Strictly Sparse Update with additional regularization to stabilize adaptive updates; (ii) a constant-corrected adaptive algorithm to attenuate the over-scaling of primitive gradients, yielding a stable mixture of adaptive and non-adaptive steps; and (iii) attributes mixture via Stochastic Attribute Averaging to mitigate frame-preference under motion disturbances. Experiments show consistent improvements and reduced generalization issues, highlighting the role of non-adaptive updates and the influence from frame-preference in explicit Dynamic GS optimization.

General Machine Learning · Hardware and Software

Chonghao Zhong, Shi Linfeng, ChenHua, Tiecheng Sun, Hao Zhao, Binhang Yuan, Chaojian Li

Training 3D Gaussian Splatting (3DGS) at billion-primitive scale is fundamentally memory-bound: each Gaussian carries a large attribute vector, and the aggregate parameter table quickly exceeds GPU capacity, limiting prior systems to tens of millions of Gaussians on consumer hardware. We observe that 3DGS training is inherently sparse and trajectory-conditioned: each iteration activates only the Gaussians visible from the current camera batch, so GPU memory can serve as a working-set cache rather than a persistent parameter store. Building on this insight, we introduce \textbf{TideGS}, an out-of-core training framework that manages parameters across an SSD--CPU--GPU hierarchy via three synergistic techniques: block-virtualized geometry for SSD-aligned spatial locality, a hierarchical asynchronous pipeline to overlap I/O with computation, and trajectory-adaptive differential streaming that transfers only incremental working-set deltas between iterations. Experiments show that TideGS enables training with \textbf{over one billion Gaussians} on a single consumer GPU while achieving state-of-the-art reconstruction quality on large-scale scenes, exceeding prior out-of-core baselines (e.g., $\sim$100M Gaussians) and standard in-memory training (e.g., $\sim$11M Gaussians).

Theory · Online Learning and Bandits

Yanlin Qu, Hongseok Namkoong, Assaf Zeevi

Thompson Sampling is one of the most widely used and studied bandit algorithms, known for its simple structure, low regret performance, and solid theoretical guarantees. Yet, in stark contrast to most other families of bandit algorithms, the exact mechanism through which posterior sampling (as introduced by Thompson) is able to "properly" balance exploration and exploitation, remains a mystery. In this paper, we show that the core insight to address this question stems from recasting Thompson Sampling as an online optimization algorithm. To distill this, we introduce a time invariant notion of regret that summarizes cumulative regret across horizons (through a regret bound), leading to a time invariant Bellman-optimal policy. It turns out that Thompson Sampling admits an online optimization form that mimics the structure of the Bellman-optimal policy, where greediness is regularized by a measure of residual uncertainty. When viewed through this new lens of online optimization, Thompson Sampling can be understood and improved in a principled manner, by comparing it against the Bellman-optimal benchmark.

General Machine Learning · Sequential, Network, and Time Series Modeling

Xiaokai Luo, Haotian Xu, Carlos Misael Madrid Padilla, OSCAR HERNAN MADRID PADILLA

We study online change point detection for multivariate inhomogeneous Poisson point process time series. This setting arises commonly in applications such as earthquake seismology, climate monitoring, and epidemic surveillance, yet remains underexplored in the machine learning and statistics literature. We propose a method that uses low-rank matrices to represent the multivariate Poisson intensity functions, resulting in an adaptive nonparametric detection procedure. Our algorithm is single-pass and requires only constant computational cost per new observation, independent of the elapsed length of the time series. We provide theoretical guarantees to control the overall false alarm probability and characterize the detection delay under temporal dependence. We also develop a new Matrix Bernstein inequality for temporally dependent Poisson point process time series, which may be of independent interest. Numerical experiments demonstrate that our method is both statistically robust and computationally efficient.