论文检索

Deep Learning · Large Language Models

Zixuan Huang, Xin Xia, Yuxi Ren, Jianbin Zheng, Xuanda Wang, Zhixia Zhang, Hongyan Xie, Songshi Liang, Zehao Chen, Xuefeng Xiao 等

Recent advancements in large reasoning models (LRMs) have greatly improved their capabilities on complex reasoning tasks through Long Chains of Thought (CoTs). However, this approach often results in substantial redundancy, impairing computational efficiency and causing significant delays in real-time applications. Recent studies show that longer reasoning chains are frequently uncorrelated with correctness and can even be detrimental to accuracy. In a further in-depth analysis of this phenomenon, we surprisingly uncover and empirically verify that LRMs implicitly know the appropriate time to stop thinking, while this capability is obscured by current sampling paradigms. Motivated by this, we introduce SAGE (Self-Aware Guided Efficient Reasoning), a novel sampling paradigm that unleashes this efficient reasoning potential. Furthermore, integrating SAGE as mixed sampling into group-based reinforcement learning (SAGE-RL) enables SAGE-RL to effectively incorporate SAGE-discovered efficient reasoning patterns into standard pass@1 inference, markedly enhancing both the reasoning accuracy and efficiency of LRMs across multiple challenging mathematical benchmarks.

Deep Learning · Robustness

Guangmingmei Yang, David Miller, George Kesidis

Most post-training backdoor detection methods rely on attacked models exhibiting extreme outlier detection statistics for the target class of an attack, compared to non-target classes. However, these approaches may fail: (1) when some (non-target) classes are easily discriminable from all others, in which case they may _naturally_ achieve extreme detection statistics (e.g., decision confidence); and (2) when the backdoor is subtle, i.e., with its features weak relative to intrinsic class-discriminative features. A key observation is that the backdoor target class has contributions to its detection statistic from both the backdoor trigger _and_ from its intrinsic features, whereas non-target classes _only_ have contributions from their intrinsic features. To achieve more sensitive detectors, we thus propose to _suppress_ intrinsic features while optimizing the detection statistic for a given class. For non-target classes, such suppression will drastically reduce the achievable statistic, whereas for the target class the (significant) contribution from the backdoor trigger remains. In practice, we formulate a constrained optimization problem, leveraging a small set of clean examples from a given class, and optimizing the detection statistic while orthogonalizing with respect to the class's intrinsic features. We dub this approach ''class subspace orthogonalization'' (CSO). CSO can be ''plug-and-play'' applied to a wide variety of existing detectors. We demonstrate its effectiveness in improving several well-known detectors, comparing with a variety of baseline detectors, against a variety of attacks, on the CIFAR-10, GTSRB, and TinyImageNet domains. Moreover, to make the detection problem even more challenging, we also evaluate against a novel mixed clean/dirty-label poisoning attack that is more surgical and harder to detect than traditional dirty-label attacks. Finally, we evaluate CSO against an adaptive attack designed to defeat it, with promising detection results.

Deep Learning · Large Language Models

Siya Qi, Yudong Chen, Runcong Zhao, Qinglin Zhu, Zhanghao Hu, Wei Liu, Yulan He, Zheng Yuan, Lin Gui

Hallucination detection is critical for ensuring the reliability of large language models (LLMs) in context-based generation. Prior work has explored intrinsic signals available during generation, among which attention offers a direct view of grounding behavior. However, existing approaches typically rely on coarse summaries that fail to capture fine-grained instabilities in attention. Inspired by signal processing, we introduce a frequency-aware perspective on attention by analyzing its variation during generation. We model attention distributions as discrete signals and extract high-frequency components that reflect rapid local changes in attention. Our analysis reveals that hallucinated tokens are associated with high-frequency attention energy, reflecting fragmented and unstable grounding behavior. Based on this insight, we develop a lightweight hallucination detector using high-frequency attention features. Experiments on the RAGTruth and HalluRAG benchmarks show that our approach achieves performance gains over verification-based, internal-representation-based, and attention-based methods across models and tasks.

Deep Learning · Generative Models and Autoencoders

Yuchen Zhu, Wei Guo, Jaemoo Choi, Petr Molodyk, Bo Yuan, Molei Tao, Yongxin Chen

Diffusion large language models (dLLMs) are promising alternatives to autoregressive large language models (AR-LLMs), as they potentially allow higher inference throughput. Reinforcement learning (RL) is a crucial component for dLLMs to achieve comparable performance with AR-LLMs on important tasks, such as reasoning. However, RL algorithms that are well-suited for dLLMs' unique characteristics have yet to be developed. This paper proposes **Distribution Matching Policy Optimization (DMPO)**, a principled and theoretically grounded RL fine-tuning method specifically designed to enhance the reasoning capabilities of dLLMs by matching the dLLM policy distribution to the optimal, reward-tilted one through cross-entropy optimization. We identify a key challenge in the implementation with a small training batch size and propose several effective solutions through a novel weight baseline subtraction technique. DMPO exhibits superior performance on multiple reasoning benchmarks without supervised fine-tuning, with an accuracy improvement of up to $54.3\\%$ over previously SOTA baselines and $66.41\\%$ over the base model, underscoring the effectiveness of the distribution matching framework.

Deep Learning · Large Language Models

Wei Liu, Siya Qi, Yali Du, Yulan He

Large language models (LLMs) make it plausible to build systems that improve through self-evolving loops, but many existing proposals are better understood as self-play and often plateau quickly. A central failure mode is that the loop synthesises more data without increasing *learnable information* for the next iteration. Through experiments on a self-play coding task, we reveal that **sustainable self-evolution requires a self-synthesised data pipeline with learnable information that increases across iterations.** We identify triadic roles that self-evolving LLMs play: the *proposer*, which generates tasks; the *solver*, which attempts solutions; and the *verifier*, which provides training signals, and we identify three system designs that jointly target learnable information gain from this triadic roles perspective. Asymmetric co-evolution closes a weak-to-strong-to-weak loop across roles. Capacity growth expands parameter and inference-time budgets to match rising learnable information. Proactive information seeking introduces external context and new task sources that prevent saturation. Together, these modules provide a measurable, system-level path from brittle self-play dynamics to sustained self-evolution.

Applications · Time Series

Aoyu Liu, Liming Wei, YAYING ZHANG

Most spatio-temporal forecasting models assume in-distribution data and can degrade sharply under non-stationary environments. Existing methods for handling distribution shift largely rely on discrete graph inference, making it difficult to disentangle universal dynamics from environment-specific changes and to respect the continuous physical nature of spatio-temporal fields. To this end, we propose STPDE, a general framework that reformulates spatio-temporal dynamics as the evolution of inhomogeneous partial differential equations. STPDE explicitly decomposes dynamics into an Invariant Diffusion Operator that captures universal mechanisms and an Environment Basis Manifold that parameterizes local heterogeneous media. We show that the Green's function of the Laplacian can be effectively approximated by linear attention, enabling global diffusion at scale. Combined with stochastic environment perturbations, STPDE improves robustness under heterogeneous and shifting environments. Extensive experiments on in-distribution forecasting, out-of-distribution generalization, few-shot cross-city transfer, and continual learning demonstrate consistent improvements over state-of-the-art baselines with competitive computational efficiency.

Deep Learning · Large Language Models

Wei Liu, Peijie Yu, Michele Orini, Yali Du, Yulan He

The agency expected of Agentic Large Language Models goes beyond answering correctly, requiring autonomy to set goals and decide what to explore. We term this *investigatory intelligence*, distinguishing it from *executional intelligence*, which merely completes assigned tasks. Data Science provides a natural testbed, as real-world analysis starts from raw data rather than explicit queries, yet few benchmarks focus on it. To address this, we introduce **Deep Data Research (DDR)**, an open-ended task where LLMs autonomously extract key insights from databases, and **DDR-Bench**, a large-scale, checklist-based benchmark that enables verifiable evaluation. Results show that while frontier models display emerging agency, long-horizon exploration remains challenging. Our analysis highlights that effective investigatory intelligence depends not only on agent scaffolding or merely scaling, but also on intrinsic strategies of agentic models.

Social Aspects · Privacy

Zhiyu Sun, Minrui Luo, Yu Wang, Tianxing He, Zhili Chen

Large language models (LLMs) are pretrained on corpora containing trillions of tokens and, therefore, inevitably memorize sensitive information. Locate-then-edit methods, as a mainstream paradigm of model editing, offer a promising solution by modifying model parameters without retraining. However, in this work, we reveal a critical vulnerability of this paradigm: the parameter updates inadvertently serve as a side channel, enabling attackers to recover the edited data. We propose a two-stage reverse-engineering attack named KSTER (KeySpaceReconsTruction-then-EntropyReduction) that leverages the low-rank structure of these updates. First, we theoretically show that the row space of the update matrix encodes a "fingerprint" of the edited subjects, enabling accurate subject recovery via spectral analysis. Second, we introduce an entropy-based prompt recovery attack that reconstructs the semantic context of the edit. Extensive experiments on multiple LLMs demonstrate that our attacks can recover edited data with high success rates. Furthermore, we propose subspace camouflage, a defense strategy that obfuscates the update fingerprint with semantic decoys. This approach effectively mitigates reconstruction risks without compromising editing utility. Our code is available at https://anonymous.4open.science/r/EditATK-26B6.

Applications · Computer Vision

Kabilan Elangovan, Daniel Ting

This position paper argues that explanation stability claims are scientifically invalid without cross-method validation. Just as statistical significance requires specifying the test statistic, stability must be validated across multiple attribution paradigms or explicitly scoped to a single method’s computational objective. In controlled chest X-ray experiments, DenseNet201, ResNet50V2, and InceptionV3 achieve >99% AUC but exhibit reversed stability rankings across attribution methods. LayerCAM ranks InceptionV3 highest (IoU 0.777), while Grad-CAM++ favors DenseNet201, reducing InceptionV3’s score by 17.3%. These findings establish that explanation stability is an emergent property of the model–method pair, not an intrinsic model trait. We call for mandatory cross-method validation in XAI research and urge that regulatory submissions specify attribution methods to avoid illusionary safety assurances.

Minrui Luo, Zhiheng Zhang

Synthetic Nearest Neighbors (SNN) provides a principled solution to causal matrix completion under missing-not-at-random (MNAR) by exploiting local low-rank structure through fully observed anchor submatrices. However, its effectiveness critically relies on sufficient data availability within each treatment level, a condition that often fails in settings with multiple or complex treatments. In this work, we propose Mixed Synthetic Nearest Neighbors (MSNN), a new entry-wise causal identification estimator that integrates information across treatment levels. We show that MSNN retains the finite-sample error bounds and asymptotic normality guarantees of SNN, while enlarging the effective sample size available for estimation. Empirical results on synthetic and real-world datasets illustrate the efficacy of the proposed approach, especially under data-scarce treatment levels.

Deep Learning · Large Language Models

Kabilan Elangovan, Jasmine Ong, Daniel Ting

This position paper argues that as Large Language Models (LLMs) increasingly consume synthetic data, parametric representations can no longer serve as reliable witnesses of factual provenance. Current architectures, which treat fluent outputs as implicitly grounded, create a critical epistemic failure mode: systems emit accurate-looking claims with no recoverable lineage to verifiable sources. We advance the position that referenceability and explicit traceability of claims to accessible evidence must be enforced as a non-negotiable system invariant. Distinct from Retrieval-Augmented Generation (RAG), which enriches generation with external context, we propose a negative safety constraint: in factual settings, no atomic claim should be emitted unless it is evidence-gated by identifiers that entail it; otherwise, the system must abstain. To operationalize this, we introduce a “separation-of-powers” architecture that decouples parametric generation from factual authorization, along with a diagnostic metric—Parametric Leakage Ratio (PLR)—to quantify ungrounded factual emissions. We conclude that enforcing a strict provenance–parametric divide is essential to prevent safety certifications from legitimizing unverifiable outputs in high-stakes domains such as healthcare.

Deep Learning · Generative Models and Autoencoders

Wenbin Duan, Yan Shu, Zhuoyuan Fu, Fangmin Zhao, Yan Li, Yaru Zhao, Binyang Li

Rectified-flow-based diffusion transformers, particularly FLUX, have demonstrated outstanding performance in high-quality image generation. However, achieving fast and accurate inversion—transforming images back to latent noise for faithful reconstruction and editing—remains a challenging bottleneck due to the discretization errors of linear solvers. This paper introduces \textbf{SlerpFlow}, a straightforward yet highly effective zero-shot approach that unlocks the full potential of FLUX for high-fidelity inversion and editing. Unlike existing approaches (e.g., RF-Solver) that rely on complex numerical approximations such as high-order Taylor expansions to correct trajectory errors, we present a geometric view based on the Manifold Hypothesis: the empirically observed trajectory curvature is not a numerical artifact, but rather serves as a necessary ``centripetal force" that constrains the flow to remain on the data manifold. Guided by this insight, SlerpFlow integrates Spherical Linear Interpolation (Slerp) to rectify flow velocity directions on the hypersphere, strictly adhering to the intrinsic curvature of the latent space. Crucially, by caching the corrected velocity for subsequent steps, SlerpFlow achieves high-precision inversion while maintaining the computational efficiency of a first-order Euler solver. Extensive experiments on FLUX-based reconstruction and editing tasks demonstrate that our method delivers superior structural consistency and lower reconstruction errors compared to state-of-the-art baselines without requiring additional training.

Deep Learning · Large Language Models

Yang Wan, Zheng Cao, Zhenhao Zhang, Zhengwen Zeng, Shuheng Shen, Changhua Meng, Linchao Zhu

Large language models excel as few-shot learners when provided with appropriate demonstrations, yet this strength becomes problematic in multi-turn agent scenarios, where LLMs erroneously mimic their own previous responses as few-shot examples. Through attention analysis, we identify \textbf{conversational inertia}, a phenomenon where models exhibit strong diagonal attention to previous responses, which is associated with imitation bias that constrains exploration. This reveals a tension when transforming few-shot LLMs into agents: longer context enriches environmental feedback for exploitation, yet also amplifies conversational inertia that undermines exploration. Our key insight is that for identical states, actions generated with longer contexts exhibit stronger inertia than those with shorter contexts, enabling construction of preference pairs without environment rewards. Based on this, we propose Context Preference Learning to calibrate model preferences to favor low-inertia responses over high-inertia ones. We further provide context management strategies at inference time to balance exploration and exploitation. Experimental results across eight agentic environments and one deep research scenario validate that our framework reduces conversational inertia and achieves performance improvements.

Deep Learning · Graph Neural Networks

Yancheng Chen, Dun Ma, Shuai Zhang, Yang Liu, Xixun Lin, Xiangyu Zhao, Wenguo Yang, Wei Chen, Chuan Zhou

Graph Foundation Models (GFMs), built upon the *Pre-training and Adaptation* paradigm, have emerged as a research hotspot in graph learning. For GNN-based GFMs, graph prompt tuning has become the prevailing adaptation method for downstream tasks. Although recent methods explain why graph prompt tuning works, how to rigorously measure its adaptation capacity remains an open problem. Addressing this problem is critical for understanding the capability limits of graph prompt tuning and for developing more powerful adaptation methods. In this paper, we propose Prismatic Space Theory (PS-Theory), a novel mathematical framework to quantify the capacity of adaptation methods, while focusing on establishing the upper bound for the adaptation capacity of graph prompt tuning. Building upon the proposed PS-Theory, we further introduce **M**essage **T**uning for **G**FMs (MTG), a lightweight approach that injects a small set of learnable message prototypes into each layer of the GNN backbone to adaptively guide message fusion without updating pre-trained weights. Through our PS-Theory, we prove that the adaptation capacity of MTG can exceed the theoretical upper bound of graph prompt tuning. Extensive experiments demonstrate that MTG consistently outperforms graph prompt baselines across diverse benchmark datasets, providing strong empirical support for our theoretical findings. Our code is available at https://anonymous.4open.science/r/MTG.

Deep Learning · Large Language Models

Jing Bi, Luchuan Song, Dingxin Zhang, Pinxin Liu, Guangyu Sun, Lianggong Wen, Weidong Cai, Chen Chen, Chenliang Xu

Vision–language models (VLMs) now support both direct Instruct and explicit-reasoning Thinking modes, but practitioners lack principled ways to decide when reasoning helps or how much computation to allocate at test time. We investigate whether VLMs encode meta-cognitive signals for adaptive inference. We derive oracle labels for two properties: (1) reasoning helpfulness—whether explicit reasoning improves accuracy, and (2) optimal generation length—the minimal token budget for correctness. Probing final-layer representations in InternVL and Qwen3-VL models, we find Thinking models encode these signals more linearly than Instruct models, suggesting reasoning-oriented training enhances meta-cognitive structure. Head-wise attribution reveals two circuits: length-control heads in lower layers, and reasoning/difficulty heads in higher layers. Causal interventions confirm these roles: scaling length heads controls output length with little accuracy loss, while scaling reasoning heads enables a perception–reasoning trade-off, improving accuracy by up to 5.3\%. These effects generalize across benchmarks. Our results show reasoning-tuned VLMs develop localized, manipulable circuits for meta-cognitive control, enabling test-time steering of computation and reasoning without retraining.

Deep Learning · Generative Models and Autoencoders

Xinbao Qiao, Xianglong Du, Wei Liu, Jingqi Zhang, Peihua Mai, Meng Zhang, Yan (James) Pang

The proliferation of recursive synthetic data training promises to alleviate data scarcity but introduces the existential risk of model collapse, wherein recursive training on synthetic data erodes distributional tails and homogenizes outputs. Current literature identifies data selection as a pivotal solution, employing verifiers to prune datasets in pursuit of synthetic samples that approximate the true data manifold. However, this approach hinges on the fragile and often unrealistic assumption that a perfect verifier possesses global distributional knowledge. In real-world scenarios characterized by data silos, such as fragmented healthcare consortia or proprietary financial institutions, this assumption is invalidated by the inherent fragmentation of knowledge. We theoretically prove that such siloed selection accelerates model collapse, driving diversity decay governed by a power law. To bridge this gap, we propose an automated filtering criterion that synergizes the sensitivity theorem with Wasserstein geometry. Specifically, multiple parties collaboratively compute geodesic interpolations and the Wasserstein Barycenter as proxy measures, without exchanging raw data. These proxies serve as a collective reference, enabling multiple parties to score synthetic data rather than relying on a single biased perspective in a data silo. Empirical results show the failure of the baseline on skewed distributions, whereas our methods effectively prevent collapse. Code available at [Anonymous Github](https://anonymous.4open.science/r/When-Sample-Selection-Bias-Precipitates-Model-Collapse-D8D4).

Applications · Chemistry, Physics, and Earth Sciences

Stefan Pranger, Bernhard Ramsauer, Oliver Hofmann, Bettina Könighofer

Scanning tunneling microscopy (STM) enables precise manipulation of individual atoms and molecules, offering a pathway to constructing nanoscale assemblies with rich quantum mechanical behavior. Despite its potential, STM-based fabrication remains limited by the inherent complexity of manipulation procedures and the extensive manual effort required. In this work, we take a substantial step toward autonomous manufacturing with STMs by introducing a novel AI-based planning framework for molecular assembly and a high-fidelity simulation environment. Our framework computes collision-free assembly plans that minimize the total distance traveled by molecules. Given an assignment of molecules to target positions, satisfiability solving is used to compute execution schedules in which each molecule has an empty corridor available when it is scheduled to move. Reinforcement learning (RL) agents then execute sequences of STM actions to manipulate molecules to their targets. We further introduce NanoAssemblyGym, a high-fidelity simulation environment for molecular manipulation built on the Gymnasium API, allowing seamless integration with existing RL libraries and workflows. Using NanoAssemblyGym, we demonstrate autonomous assembly of structures containing up to 420 molecules.

Deep Learning · Attention Mechanisms

Zeru Shi, Zhenting Wang, Fan Yang, Qifan Wang, Ruixiang Tang

We investigate the origins of massive activations in large language models (LLMs) and identify a specific layer named the Massive Emergence Layer (ME Layer), that is consistently observed across model families, where massive activations first emerge and subsequently propagate to deeper layers through residual connections. We show that, within the ME Layer both the RMSNorm and the FFN parameters jointly contribute to the emergence of massive activations. Once formed, the massive activation token representation remains largely invariant across layers, reducing the diversity of hidden representations passed to the attention module. Motivated by this limitation, we propose a simple and effective method to reduce the rigidity of the massive activation token. Our approach consistently improves LLM performance across multiple tasks, including instruction following and math reasoning, in both training free and fine tuning settings. Moreover, we show that our method mitigates attention sinks by selectively weakening their influence, elucidating their origin at the hidden state level and shedding new light on principled mitigation strategies.

Social Aspects · Everything Else

Ziming Hong, Tianyu Huang, Runnan Chen, Shanshan Ye, Mingming Gong, Bo Han, Tongliang Liu

Recent studies have extended diffusion-based instruction-driven 2D image editing pipelines to 3D Gaussian Splatting (3DGS), enabling faithful manipulation of 3DGS assets and greatly advancing 3DGS content creation. However, it also exposes these assets to serious risks of unauthorized editing and malicious tampering. Although imperceptible adversarial perturbations against diffusion models have proven effective for protecting 2D images, applying them to 3DGS encounters two major challenges: view-generalizable protection and balancing invisibility with protection capability. In this work, we propose the first editing safeguard for 3DGS, termed AdLift, which prevents instruction-driven editing across arbitrary views and dimensions by lifting strictly bounded 2D adversarial perturbations into 3D Gaussian-represented safeguard. To ensure both adversarial perturbations effectiveness and invisibility, these safeguard Gaussians are progressively optimized across training views using a tailored Lifted PGD, which first conducts gradient truncation during back-propagation from the editing model at the rendered image and applies projected gradients to strictly constrain the image-level perturbation. Then, the resulting perturbation is backpropagated to the safeguard Gaussian parameters via an image-to-Gaussian fitting operation. We alternate between gradient truncation and image-to-Gaussian fitting, yielding consistent adversarial-based protection performance across different viewpoints and generalizes to novel views. Empirically, qualitative and quantitative results demonstrate that AdLift effectively protects against state-of-the-art instruction-driven 2D image and 3DGS editing.

Applications · Time Series

Heng Zhou, Xin Sun, Chao LI

Real-world time series inherently exhibit complex temporal patterns. Within chaotic systems, significant mixing and entanglement occur between different time-varying modes. Given that time series exhibit distinctly different patterns at various sampling scales, downsampling to extract multiscale features is a common approach. However, conventional downsampling causes high-frequency components in the original signal, those exceeding the new Nyquist frequency, to undergo spectral folding. This erroneously introduces spurious low-frequency patterns, perceived as low-frequency noise, thereby leading to the ***aliasing problem***. To address this problem, we propose a Decomposition-Prevention-Fusion architecture framework called **DMANet**, which introduces the **D**ynamic **M**ultiscale **A**nti-Aliasing **Net**work. Specifically, DMANet comprises two key components: Multiscale Convolutional Downsampling, designed to capture temporal dependencies and inter-channel interactions, and an Anti-Aliasing Operation, which includes Pre-Sampling Anti-Aliasing Filtering and Post-Sampling Interpolation. These designs guarantee the fidelity of multiscale features before and after downsampling. We show that by mitigating the risk of aliasing, our proposed simple convolutional downsampling architecture achieves performance competitive with common baselines. Our codes are available at https://anonymous.4open.science/r/DMANet.