# Research Papers — 2025-12 · bundle page 133 of 175 > Full-text markdown of papers #6601–6650 of 8703 published in 2025-12, newest first. AI-readable mirrors of the full 8703-paper 2025-12 archive (50 papers per file). Canonical page: https://aipod.dev/papers/archive/2025-12/page/133 [← Previous bundle](https://aipod.dev/papers/archive/2025-12/page/132.md) [Next bundle →](https://aipod.dev/papers/archive/2025-12/page/134.md) ### Enhancing Interpretability of AR-SSVEP-Based Motor Intention Recognition via CNN-BiLSTM and SHAP Analysis on EEG Data Authors: Lin Yang, Xiang Li, Xin Ma, Xinxin Zhao Published: 2025-12-07 Categories: cs.LG, cs.CV arXiv: https://arxiv.org/abs/2512.06730v1 Patients with motor dysfunction show low subjective engagement in rehabilitation training. Traditional SSVEP-based brain-computer interface (BCI) systems rely heavily on external visual stimulus equipment, limiting their practicality in real-world settings. This study proposes an augmented reality steady-state visually evoked potential (AR-SSVEP) system to address the lack of patient initiative and the high workload on therapists. Firstly, we design four HoloLens 2-based EEG classes and collect EEG data from seven healthy subjects for analysis. Secondly, we build upon the conventional CNN-BiLSTM architecture by integrating a multi-head attention mechanism (MACNN-BiLSTM). We extract ten temporal-spectral EEG features and feed them into a CNN to learn high-level representations. Then, we use BiLSTM to model sequential dependencies and apply a multi-head attention mechanism to highlight motor-intention-related patterns. Finally, the SHAP (SHapley Additive exPlanations) method is applied to visualize EEG feature contributions to the neural network's decision-making process, enhancing the model's interpretability. These findings enhance real-time motor intention recognition and support recovery in patients with motor impairments. --- ### KV-CAR: KV Cache Compression using Autoencoders and KV Reuse in Large Language Models Authors: Sourjya Roy, Shrihari Sridharan, Surya Selvam, Anand Raghunathan Published: 2025-12-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.06727v1 As Large Language Models (LLMs) scale in size and context length, the memory requirements of the key value (KV) cache have emerged as a major bottleneck during autoregressive decoding. The KV cache grows with sequence length and embedding dimension, often exceeding the memory footprint of the model itself and limiting achievable batch sizes and context windows. To address this challenge, we present KV CAR, a unified and architecture agnostic framework that significantly reduces KV cache storage while maintaining model fidelity. KV CAR combines two complementary techniques. First, a lightweight autoencoder learns compact representations of key and value tensors along the embedding dimension, compressing them before they are stored in the KV cache and restoring them upon retrieval. Second, a similarity driven reuse mechanism identifies opportunities to reuse KV tensors of specific attention heads across adjacent layers. Together, these methods reduce the dimensional and structural redundancy in KV tensors without requiring changes to the transformer architecture. Evaluations on GPT 2 and TinyLLaMA models across Wikitext, C4, PIQA, and Winogrande datasets demonstrate that KV CAR achieves up to 47.85 percent KV cache memory reduction with minimal impact on perplexity and zero shot accuracy. System level measurements on an NVIDIA A40 GPU show that the reduced KV footprint directly translates into longer sequence lengths and larger batch sizes during inference. These results highlight the effectiveness of KV CAR in enabling memory efficient LLM inference. --- ### The Role of Entropy in Visual Grounding: Analysis and Optimization Authors: Shuo Li, Jiajun Sun, Zhihao Zhang, Xiaoran Fan, Senjie Jin, Hui Li, Yuming Yang, Junjie Ye, Lixing Shen, Tao Ji, Tao Gui, Qi Zhang et al. Published: 2025-12-07 Categories: cs.CV, cs.AI, cs.CL arXiv: https://arxiv.org/abs/2512.06726v1 Recent advances in fine-tuning multimodal large language models (MLLMs) using reinforcement learning have achieved remarkable progress, particularly with the introduction of various entropy control techniques. However, the role and characteristics of entropy in perception-oriented tasks like visual grounding, as well as effective strategies for controlling it, remain largely unexplored. To address this issue, we focus on the visual grounding task and analyze the role and characteristics of entropy in comparison to reasoning tasks. Building on these findings, we introduce ECVGPO (Entropy Control Visual Grounding Policy Optimization), an interpretable algorithm designed for effective entropy regulation. Through entropy control, the trade-off between exploration and exploitation is better balanced. Experiments show that ECVGPO achieves broad improvements across various benchmarks and models. --- ### Decoding Motor Behavior Using Deep Learning and Reservoir Computing Authors: Tian Lan Published: 2025-12-07 Categories: cs.LG, eess.SP arXiv: https://arxiv.org/abs/2512.06725v1 We present a novel approach to EEG decoding for non-invasive brain machine interfaces (BMIs), with a focus on motor-behavior classification. While conventional convolutional architectures such as EEGNet and DeepConvNet are effective in capturing local spatial patterns, they are markedly less suited for modeling long-range temporal dependencies and nonlinear dynamics. To address this limitation, we integrate an Echo State Network (ESN), a prominent paradigm in reservoir computing into the decoding pipeline. ESNs construct a high-dimensional, sparsely connected recurrent reservoir that excels at tracking temporal dynamics, thereby complementing the spatial representational power of CNNs. Evaluated on a skateboard-trick EEG dataset preprocessed via the PREP pipeline and implemented in MNE-Python, our ESNNet achieves 83.2% within-subject and 51.3% LOSO accuracies, surpassing widely used CNN-based baselines. Code is available at https://github.com/Yutiankunkun/Motion-Decoding-Using-Biosignals --- ### ProAgent: Harnessing On-Demand Sensory Contexts for Proactive LLM Agent Systems in the Wild Authors: Bufang Yang, Lilin Xu, Liekang Zeng, Yunqi Guo, Siyang Jiang, Wenrui Lu, Kaiwei Liu, Yixuan Li, Xiaofan Jiang, Guoliang Xing, Zhenyu Yan Published: 2025-12-07 Categories: cs.AI, cs.CL, cs.HC arXiv: https://arxiv.org/abs/2512.06721v2 Recent studies have begun to explore proactive large language model (LLM) agents that provide unobtrusive assistance by automatically leveraging contextual information, such as in code editing and in-app suggestions. However, most focus on short, task-specific episodes or on-screen contexts, rather than continuously perceiving and assisting users throughout daily life. Enabling such in-the-wild assistance requires continuous sensing of users' surroundings, which can incur substantial system overhead. In this work, we propose ProAgent, an end-to-end proactive agent system that harnesses on-demand sensory contexts to provide in-the-wild assistance. ProAgent first employs on-demand tiered perception to continuously sense users' surroundings by integrating low-cost contextual cues with richer perception on demand, and uses proactive-oriented context extraction to derive hierarchical contexts integrating both sensory contexts and human preferences. ProAgent then employs a context-aware proactive reasoner to infer user needs and invokes external tools to deliver proactive assistance. We implement ProAgent on AR glasses and evaluate it on a public dataset and a real-world dataset. Results demonstrate that ProAgent achieves up to 27.7% higher proactive prediction accuracy and 20.5% lower false detection than state-of-the-art baselines. A user study with 20 participants shows that 85% were satisfied with ProAgent and willing to use it in daily life. --- ### ProAgent: Harnessing On-Demand Sensory Contexts for Proactive LLM Agent Systems Authors: Bufang Yang, Lilin Xu, Liekang Zeng, Yunqi Guo, Siyang Jiang, Wenrui Lu, Kaiwei Liu, Hancheng Xiang, Xiaofan Jiang, Guoliang Xing, Zhenyu Yan Published: 2025-12-07 Categories: cs.AI, cs.CL, cs.HC arXiv: https://arxiv.org/abs/2512.06721v1 Large Language Model (LLM) agents are emerging to transform daily life. However, existing LLM agents primarily follow a reactive paradigm, relying on explicit user instructions to initiate services, which increases both physical and cognitive workload. In this paper, we propose ProAgent, the first end-to-end proactive agent system that harnesses massive sensory contexts and LLM reasoning to deliver proactive assistance. ProAgent first employs a proactive-oriented context extraction approach with on-demand tiered perception to continuously sense the environment and derive hierarchical contexts that incorporate both sensory and persona cues. ProAgent then adopts a context-aware proactive reasoner to map these contexts to user needs and tool calls, providing proactive assistance. We implement ProAgent on Augmented Reality (AR) glasses with an edge server and extensively evaluate it on a real-world testbed, a public dataset, and through a user study. Results show that ProAgent achieves up to 33.4% higher proactive prediction accuracy, 16.8% higher tool-calling F1 score, and notable improvements in user satisfaction over state-of-the-art baselines, marking a significant step toward proactive assistants. A video demonstration of ProAgent is available at https://youtu.be/pRXZuzvrcVs. --- ### SIEVE: Selective Integrity Verification and Escalation for Defending LLM Agents against Indirect Prompt Injection Authors: Zhibo Liang, Tianze Hu, Zaiye Chen, Mingjie Tang Published: 2025-12-07 Categories: cs.AI, cs.CL, cs.CR arXiv: https://arxiv.org/abs/2512.06716v3 Large Language Models (LLMs) are increasingly used as the core of agentic systems due to their strong reasoning, planning, and tool-use capabilities. By interacting with external environments, LLM agents can execute real-world tasks on behalf of users rather than merely generate text. This expanded capability also amplifies the threat of indirect prompt injection (IPI), where malicious external content can manipulate agent behavior and trigger unauthorized actions, privacy leakage, or financial loss. Existing defenses generally follow two approaches. Plan- or rule-based methods constrain agent execution using predefined plans or execution rules, but may block legitimate actions that arise from dynamic runtime context. Semantic auditing methods offer greater flexibility, yet repeatedly re-evaluating proposed actions incurs substantial token and latency overhead. These limitations motivate a selective verification strategy that applies deterministic checks whenever reliable structural evidence is available and escalates only ambiguous cases. Accordingly, we propose SIEVE, which verifies tool transitions and argument sources against an Intent Graph, escalating actions that cannot pass deterministic verification to semantic adjudication. This selective design preserves flexibility while avoiding continuous semantic auditing. We evaluate SIEVE on AgentLure and AgentDojo against MELON, DRIFT, IPIGuard, and ARGUS. SIEVE achieves 5.94% ASR with 97.5% clean utility on AgentLure, and 0.34% ASR with 87.63% clean utility on AgentDojo. Compared with DRIFT and ARGUS, SIEVE incurs the lowest token consumption and the fewest API calls. --- ### Cognitive Control Architecture (CCA): A Lifecycle Supervision Framework for Robustly Aligned AI Agents Authors: Zhibo Liang, Tianze Hu, Zaiye Chen, Mingjie Tang Published: 2025-12-07 Categories: cs.AI, cs.CL, cs.CR arXiv: https://arxiv.org/abs/2512.06716v1 Autonomous Large Language Model (LLM) agents exhibit significant vulnerability to Indirect Prompt Injection (IPI) attacks. These attacks hijack agent behavior by polluting external information sources, exploiting fundamental trade-offs between security and functionality in existing defense mechanisms. This leads to malicious and unauthorized tool invocations, diverting agents from their original objectives. The success of complex IPIs reveals a deeper systemic fragility: while current defenses demonstrate some effectiveness, most defense architectures are inherently fragmented. Consequently, they fail to provide full integrity assurance across the entire task execution pipeline, forcing unacceptable multi-dimensional compromises among security, functionality, and efficiency. Our method is predicated on a core insight: no matter how subtle an IPI attack, its pursuit of a malicious objective will ultimately manifest as a detectable deviation in the action trajectory, distinct from the expected legitimate plan. Based on this, we propose the Cognitive Control Architecture (CCA), a holistic framework achieving full-lifecycle cognitive supervision. CCA constructs an efficient, dual-layered defense system through two synergistic pillars: (i) proactive and preemptive control-flow and data-flow integrity enforcement via a pre-generated "Intent Graph"; and (ii) an innovative "Tiered Adjudicator" that, upon deviation detection, initiates deep reasoning based on multi-dimensional scoring, specifically designed to counter complex conditional attacks. Experiments on the AgentDojo benchmark substantiate that CCA not only effectively withstands sophisticated attacks that challenge other advanced defense methods but also achieves uncompromised security with notable efficiency and robustness, thereby reconciling the aforementioned multi-dimensional trade-off. --- ### A Novel Deep Neural Network Architecture for Real-Time Water Demand Forecasting Authors: Tony Salloom, Okyay Kaynak, Wei He Published: 2025-12-07 Categories: cs.LG, cs.AI, cs.RO arXiv: https://arxiv.org/abs/2512.06714v1 Short-term water demand forecasting (StWDF) is the foundation stone in the derivation of an optimal plan for controlling water supply systems. Deep learning (DL) approaches provide the most accurate solutions for this purpose. However, they suffer from complexity problem due to the massive number of parameters, in addition to the high forecasting error at the extreme points. In this work, an effective method to alleviate the error at these points is proposed. It is based on extending the data by inserting virtual data within the actual data to relieve the nonlinearity around them. To our knowledge, this is the first work that considers the problem related to the extreme points. Moreover, the water demand forecasting model proposed in this work is a novel DL model with relatively low complexity. The basic model uses the gated recurrent unit (GRU) to handle the sequential relationship in the historical demand data, while an unsupervised classification method, K-means, is introduced for the creation of new features to enhance the prediction accuracy with less number of parameters. Real data obtained from two different water plants in China are used to train and verify the model proposed. The prediction results and the comparison with the state-of-the-art illustrate that the method proposed reduces the complexity of the model six times of what achieved in the literature while conserving the same accuracy. Furthermore, it is found that extending the data set significantly reduces the error by about 30%. However, it increases the training time. --- ### Look Twice before You Leap: A Rational Framework for Localized Adversarial Anonymization Authors: Donghang Duan, Xu Zheng, Yuefeng He, Chong Mu, Leyi Cai, Lizong Zhang Published: 2025-12-07 Categories: cs.CR, cs.CL arXiv: https://arxiv.org/abs/2512.06713v3 Current LLM-based frameworks for text anonymization usually rely on remote API services from powerful LLMs, which creates an inherent privacy paradox: users must disclose the raw data to untrusted third parties for guaranteed privacy preservation. Moreover, directly migrating current solutions to local small-scale models (LSMs) offers a suboptimal solution with severe utility collapse. Our work argues that this failure stems not merely from the capability deficits of LSMs, but significantly from the inherent irrationality of the greedy adversarial strategies employed by current state-of-the-art (SOTA) methods. To address this drawback, we propose Rational Localized Adversarial Anonymization (RLAA), a fully localized and training-free framework featuring an Attacker-Arbitrator-Anonymizer architecture. We model the anonymization process as a trade-off between Marginal Privacy Gain (MPG) and Marginal Utility Cost (MUC), demonstrating that greedy strategies tend to drift into an irrational state. Instead, RLAA introduces an arbitrator that acts as a rationality gatekeeper, validating the attacker's inference to filter out ghost leaks. This mechanism promotes a rational early-stopping criterion, and structurally prevents utility collapse. Extensive experiments on different benchmarks demonstrate that RLAA achieves a superior privacy-utility trade-off compared to strong baselines. --- ### Look Twice before You Leap: A Rational Agent Framework for Localized Adversarial Anonymization Authors: Donghang Duan, Xu Zheng Published: 2025-12-07 Categories: cs.CR, cs.CL arXiv: https://arxiv.org/abs/2512.06713v1 Current LLM-based text anonymization frameworks usually rely on remote API services from powerful LLMs, which creates an inherent "privacy paradox": users must somehow disclose data to untrusted third parties for superior privacy preservation. Moreover, directly migrating these frameworks to local small-scale models (LSMs) offers a suboptimal solution with catastrophic collapse in utility based on our core findings. Our work argues that this failure stems not merely from the capability deficits of LSMs, but from the inherent irrationality of the greedy adversarial strategies employed by current state-of-the-art (SoTA) methods. We model the anonymization process as a trade-off between Marginal Privacy Gain (MPG) and Marginal Utility Cost (MUC), and demonstrate that greedy strategies inevitably drift into an irrational state. To address this, we propose Rational Localized Adversarial Anonymization (RLAA), a fully localized and training-free framework featuring an Attacker-Arbitrator-Anonymizer (A-A-A) architecture. RLAA introduces an arbitrator that acts as a rationality gatekeeper, validating the attacker's inference to filter out feedback providing negligible benefits on privacy preservation. This mechanism enforces a rational early-stopping criterion, and systematically prevents utility collapse. Extensive experiments on different datasets demonstrate that RLAA achieves the best privacy-utility trade-off, and in some cases even outperforms SoTA on the Pareto principle. Our code and datasets will be released upon acceptance. --- ### Parameter-Efficient Fine-Tuning with Differential Privacy for Robust Instruction Adaptation in Large Language Models Authors: Yulin Huang, Yaxuan Luan, Jinxu Guo, Xiangchen Song, Yuchen Liu Published: 2025-12-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.06711v1 This study addresses the issues of privacy protection and efficiency in instruction fine-tuning of large-scale language models by proposing a parameter-efficient method that integrates differential privacy noise allocation with gradient clipping in a collaborative optimization framework. The method keeps the backbone model frozen and updates parameters through a low-dimensional projection subspace, while introducing clipping and adaptive noise allocation during gradient computation. This design reduces privacy budget consumption and ensures training stability and robustness. The unified framework combines gradient constraints, noise allocation, and parameter projection, effectively mitigating performance fluctuations and privacy risks in multi-task instruction scenarios. Experiments are conducted across hyperparameter, environment, and data sensitivity dimensions. Results show that the method outperforms baseline models in accuracy, privacy budget, and parameter efficiency, and maintains stable performance under diverse and uncertain data conditions. The findings enrich the theoretical integration of differential privacy and parameter-efficient fine-tuning and demonstrate its practical adaptability in instruction tasks, providing a feasible solution for secure training in complex instruction environments. --- ### Stochasticity in Agentic Evaluations: Quantifying Inconsistency with Intraclass Correlation Authors: Zairah Mustahsan, Abel Lim, Megna Anand, Saahil Jain, Bryan McCann Published: 2025-12-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.06710v1 As large language models become components of larger agentic systems, evaluation reliability becomes critical: unreliable sub-agents introduce brittleness into downstream system behavior. Yet current evaluation practice, reporting a single accuracy number from a single run, obscures the variance underlying these results, making it impossible to distinguish genuine capability improvements from lucky sampling. We propose adopting Intraclass Correlation Coefficient (ICC), a metric from measurement science, to characterize this variance. ICC decomposes observed variance into between-query variance (task difficulty) and within-query variance (agent inconsistency), highlighting whether reported results reflect true capability or measurement noise. We evaluated on GAIA (Levels 1-3, measuring agentic capabilities across varying reasoning complexity) and FRAMES (measuring retrieval and factuality across multiple documents). We found that ICC varies dramatically with task structure, with reasoning and retrieval tasks (FRAMES) exhibit ICC=0.4955-0.7118 across models, and agentic tasks (GAIA) exhibiting ICC=0.304-0.774 across models. For sub-agent replacement decisions in agentic systems, accuracy improvements are only trustworthy if ICC also improves. We demonstrate that ICC converges by n=8-16 trials for structured tasks and n>=32 for complex reasoning, enabling practitioners to set evidence-based resampling budgets. We recommend reporting accuracy alongside ICC and within-query variance as standard practice, and propose updated Evaluation Cards capturing these metrics. By making evaluation stability visible, we aim to transform agentic benchmarking from opaque leaderboard competition to trustworthy experimental science. Our code is open-sourced at https://github.com/youdotcom-oss/stochastic-agent-evals. --- ### A Novel Multimodal RUL Framework for Remaining Useful Life Estimation with Layer-wise Explanations Authors: Waleed Razzaq, Yun-Bo Zhao Published: 2025-12-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.06708v1 Estimating the Remaining Useful Life (RUL) of mechanical systems is pivotal in Prognostics and Health Management (PHM). Rolling-element bearings are among the most frequent causes of machinery failure, highlighting the need for robust RUL estimation methods. Existing approaches often suffer from poor generalization, lack of robustness, high data demands, and limited interpretability. This paper proposes a novel multimodal-RUL framework that jointly leverages image representations (ImR) and time-frequency representations (TFR) of multichannel, nonstationary vibration signals. The architecture comprises three branches: (1) an ImR branch and (2) a TFR branch, both employing multiple dilated convolutional blocks with residual connections to extract spatial degradation features; and (3) a fusion branch that concatenates these features and feeds them into an LSTM to model temporal degradation patterns. A multi-head attention mechanism subsequently emphasizes salient features, followed by linear layers for final RUL regression. To enable effective multimodal learning, vibration signals are converted into ImR via the Bresenham line algorithm and into TFR using Continuous Wavelet Transform. We also introduce multimodal Layer-wise Relevance Propagation (multimodal-LRP), a tailored explainability technique that significantly enhances model transparency. The approach is validated on the XJTU-SY and PRONOSTIA benchmark datasets. Results show that our method matches or surpasses state-of-the-art baselines under both seen and unseen operating conditions, while requiring ~28 % less training data on XJTU-SY and ~48 % less on PRONOSTIA. The model exhibits strong noise resilience, and multimodal-LRP visualizations confirm the interpretability and trustworthiness of predictions, making the framework highly suitable for real-world industrial deployment. --- ### Academic journals' AI policies fail to curb the surge in AI-assisted academic writing Authors: Yongyuan He, Yi Bu Published: 2025-12-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.06705v2 The rapid integration of generative AI into academic writing has prompted widespread policy responses from journals and publishers. However, the effectiveness of these policies remains unclear. Here, we analyze 5,114 journals and over 5.2 million papers to evaluate the real-world impact of AI usage guidelines. We show that despite 70% of journals adopting AI policies (primarily requiring disclosure), researchers' use of AI writing tools has increased dramatically across disciplines, with no significant difference between journals with or without policies. Non-English-speaking countries, physical sciences, and high-OA journals exhibit the highest growth rates. Crucially, full-text analysis on 164k scientific publications reveals a striking transparency gap: Of the 75k papers published since 2023, only 76 (~0.1%) explicitly disclosed AI use. Our findings suggest that current policies have largely failed to promote transparency or restrain AI adoption. We urge a re-evaluation of ethical frameworks to foster responsible AI integration in science. --- ### Pathway to $O(\sqrt{d})$ Complexity bound under Wasserstein metric of flow-based models Authors: Xiangjun Meng, Zhongjian Wang Published: 2025-12-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.06702v1 We provide attainable analytical tools to estimate the error of flow-based generative models under the Wasserstein metric and to establish the optimal sampling iteration complexity bound with respect to dimension as $O(\sqrt{d})$. We show this error can be explicitly controlled by two parts: the Lipschitzness of the push-forward maps of the backward flow which scales independently of the dimension; and a local discretization error scales $O(\sqrt{d})$ in terms of dimension. The former one is related to the existence of Lipschitz changes of variables induced by the (heat) flow. The latter one consists of the regularity of the score function in both spatial and temporal directions. These assumptions are valid in the flow-based generative model associated with the Föllmer process and $1$-rectified flow under the Gaussian tail assumption. As a consequence, we show that the sampling iteration complexity grows linearly with the square root of the trace of the covariance operator, which is related to the invariant distribution of the forward process. --- ### Predictive Modeling of I/O Performance for Machine Learning Training Pipelines: A Data-Driven Approach to Storage Optimization Authors: Karthik Prabhakar, Durgamadhab Mishra Published: 2025-12-07 Categories: cs.PF, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2512.06699v2 Modern machine learning training is increasingly bottlenecked by data I/O rather than compute. GPUs often sit idle at below 50% utilization waiting for data. This paper presents a machine learning approach to predict I/O performance and recommend optimal storage configurations for ML training pipelines. We collected 141 observations through systematic benchmarking across different storage backends (NVMe SSD, network-attached storage, in-memory filesystems), data formats, and access patterns, covering both low-level I/O operations and full training pipelines. After evaluating seven regression models and three classification approaches, XGBoost achieved the best performance with R-squared of 0.991, predicting I/O throughput within 11.8% error on average. Feature importance analysis revealed that throughput metrics and batch size are the primary performance drivers. This data-driven approach can reduce configuration time from days of trial-and-error to minutes of predictive recommendation. The methodology is reproducible and extensible to other resource management problems in ML systems. Code and data are available at https://github.com/knkarthik01/gpu_storage_ml_project --- ### Mitigating Barren Plateaus in Quantum Denoising Diffusion Probabilistic Model Authors: Haipeng Cao, Kaining Zhang, Dacheng Tao, Zhaofeng Su Published: 2025-12-07 Categories: cs.LG, quant-ph arXiv: https://arxiv.org/abs/2512.06695v3 Quantum generative models exploit quantum superposition and entanglement to enhance learning efficiency for both classical and quantum data. Recently, inspired by classical diffusion frameworks, the quantum denoising diffusion probabilistic model has emerged as a powerful tool for learning correlated noise models, many-body phases, and topological data structures. However, we demonstrate that this framework is currently restricted to small-scale systems. As the system size increases, a severe barren plateau problem emerges, fundamentally limiting the model's scalability. We provide rigorous theoretical proofs and experimental validation to identify the origin of this barren plateau, distinct from previously known causes. To restore trainability, we introduce an enhanced architecture that effectively mitigates the barren plateau phenomenon and guarantees the model's trainability in the tested settings. Building on this architecture, we further propose a conditional quantum denoising diffusion probabilistic model, capable of generating ground states based on Hamiltonian parameters, expanding the utility of quantum generative models for complex quantum state preparation to a certain extent. Our approach not only holds the potential to address the scalability and trainability bottlenecks of quantum diffusion models, but also provides a robust tool for exploring complex quantum matter and state preparation in the NISQ era. --- ### TopiCLEAR: Adaptive embedding clustering for interpretable topic discovery from short texts Authors: Aoi Fujita, Taichi Yamamoto, Yuri Nakayama, Ryota Kobayashi Published: 2025-12-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.06694v2 Topic discovery is a fundamental technique for text mining that identifies abstract topics within large document collections. A recent approach to topic discovery is to cluster document or sentence embeddings, typically obtained from pre-trained language models, and represent each cluster as a topic. Despite their strong empirical performance, the design principles linking the geometry of embedding spaces to human-interpretable topic organization remain unclear. Clarifying the relationship between these geometric structures and topic interpretability is therefore a key challenge in topic discovery. In this study, we propose TopiCLEAR (Topic discovery by CLustering Embeddings with Adaptive dimensionality Reduction), a simple framework that integrates document embeddings with iterative clustering based on adaptive dimensionality reduction. TopiCLEAR is guided by the hypothesis that human-interpretable topics correspond to low-dimensional geometric structures in embedding spaces and leverages adaptive dimensionality reduction to identify them. We evaluate topic quality using a document-level approach that combines quantitative evaluation based on human-labeled data with qualitative assessment in the absence of ground-truth labels. Experiments on four benchmark datasets, covering both formal and informal texts, show that TopiCLEAR consistently achieves strong agreement with human annotations, particularly for short and informal texts. Furthermore, a case study on Twitter data demonstrates that TopiCLEAR produces more interpretable topics than Latent Dirichlet Allocation (LDA), recovering both human-annotated topic structure and coherent sub-topic structure. These results highlight the effectiveness of clustering documents in a low-dimensional topic space for topic discovery. --- ### State Diversity Matters in Offline Behavior Distillation Authors: Shiye Lei, Zhihao Cheng, Dacheng Tao Published: 2025-12-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.06692v1 Offline Behavior Distillation (OBD), which condenses massive offline RL data into a compact synthetic behavioral dataset, offers a promising approach for efficient policy training and can be applied across various downstream RL tasks. In this paper, we uncover a misalignment between original and distilled datasets, observing that a high-quality original dataset does not necessarily yield a superior synthetic dataset. Through an empirical analysis of policy performance under varying levels of training loss, we show that datasets with greater state diversity outperforms those with higher state quality when training loss is substantial, as is often the case in OBD, whereas the relationship reverses under minimal loss, which contributes to the misalignment. By associating state quality and diversity in reducing pivotal and surrounding error, respectively, our theoretical analysis establishes that surrounding error plays a more crucial role in policy performance when pivotal error is large, thereby highlighting the importance of state diversity in OBD scenario. Furthermore, we propose a novel yet simple algorithm, state density weighted (SDW) OBD, which emphasizes state diversity by weighting the distillation objective using the reciprocal of state density, thereby distilling a more diverse state information into synthetic data. Extensive experiments across multiple D4RL datasets confirm that SDW significantly enhances OBD performance when the original dataset exhibits limited state diversity. --- ### Think-While-Generating: On-the-Fly Reasoning for Personalized Long-Form Generation Authors: Chengbing Wang, Yang Zhang, Wenjie Wang, Xiaoyan Zhao, Fuli Feng, Xiangnan He, Tat-Seng Chua Published: 2025-12-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.06690v3 Preference alignment has enabled large language models (LLMs) to better reflect human expectations, but current methods mostly optimize for population-level preferences, overlooking individual users. Personalization is essential, yet early approaches-such as prompt customization or fine-tuning-struggle to reason over implicit preferences, limiting real-world effectiveness. Recent "think-then-generate" methods address this by reasoning before response generation. However, they face challenges in long-form generation: their static one-shot reasoning must capture all relevant information for the full response generation, making learning difficult and limiting adaptability to evolving content. To address this issue, we propose FlyThinker, an efficient "think-while-generating" framework for personalized long-form generation. FlyThinker employs a separate reasoning model that generates latent token-level reasoning in parallel, which is fused into the generation model to dynamically guide response generation. This design enables reasoning and generation to run concurrently, ensuring inference efficiency. In addition, the reasoning model is designed to depend only on previous responses rather than its own prior outputs, which preserves training parallelism across different positions-allowing all reasoning tokens for training data to be produced in a single forward pass like standard LLM training, ensuring training efficiency. Extensive experiments on real-world benchmarks demonstrate that FlyThinker achieves better personalized generation while keeping training and inference efficiency. Our code is available at https://github.com/wcb0219-sketch/FlyThinker.git. --- ### Lightweight Wasserstein Audio-Visual Model for Unified Speech Enhancement and Separation Authors: Jisoo Park, Seonghak Lee, Guisik Kim, Taewoo Kim, Junseok Kwon Published: 2025-12-07 Categories: cs.CV, eess.AS arXiv: https://arxiv.org/abs/2512.06689v1 Speech Enhancement (SE) and Speech Separation (SS) have traditionally been treated as distinct tasks in speech processing. However, real-world audio often involves both background noise and overlapping speakers, motivating the need for a unified solution. While recent approaches have attempted to integrate SE and SS within multi-stage architectures, these approaches typically involve complex, parameter-heavy models and rely on supervised training, limiting scalability and generalization. In this work, we propose UniVoiceLite, a lightweight and unsupervised audio-visual framework that unifies SE and SS within a single model. UniVoiceLite leverages lip motion and facial identity cues to guide speech extraction and employs Wasserstein distance regularization to stabilize the latent space without requiring paired noisy-clean data. Experimental results demonstrate that UniVoiceLite achieves strong performance in both noisy and multi-speaker scenarios, combining efficiency with robust generalization. The source code is available at https://github.com/jisoo-o/UniVoiceLite. --- ### PersonaMem-v2: Towards Personalized Intelligence via Learning Implicit User Personas and Agentic Memory Authors: Bowen Jiang, Yuan Yuan, Maohao Shen, Zhuoqun Hao, Zhangchen Xu, Zichen Chen, Ziyi Liu, Anvesh Rao Vijjini, Jiashu He, Hanchao Yu, Radha Poovendran, Gregory Wornell et al. Published: 2025-12-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.06688v1 Personalization is one of the next milestones in advancing AI capability and alignment. We introduce PersonaMem-v2, the state-of-the-art dataset for LLM personalization that simulates 1,000 realistic user-chatbot interactions on 300+ scenarios, 20,000+ user preferences, and 128k-token context windows, where most user preferences are implicitly revealed to reflect real-world interactions. Using this data, we investigate how reinforcement fine-tuning enables a model to improve its long-context reasoning capabilities for user understanding and personalization. We also develop a framework for training an agentic memory system, which maintains a single, human-readable memory that grows with each user over time. In our experiments, frontier LLMs still struggle with implicit personalization, achieving only 37-48% accuracy. While they support long context windows, reasoning remains the bottleneck for implicit personalization tasks. Using reinforcement fine-tuning, we successfully train Qwen3-4B to outperforms GPT-5, reaching 53% accuracy in implicit personalization. Moreover, our agentic memory framework achieves state-of-the-art 55% accuracy while using 16x fewer input tokens, relying on a 2k-token memory instead of full 32k conversation histories. These results underscore the impact of our dataset and demonstrate agentic memory as a scalable path toward real-world personalized intelligence. --- ### EMGauss: Continuous Slice-to-3D Reconstruction via Dynamic Gaussian Modeling in Volume Electron Microscopy Authors: Yumeng He, Zanwei Zhou, Yekun Zheng, Chen Liang, Yunbo Wang, Xiaokang Yang Published: 2025-12-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.06684v2 Volume electron microscopy (vEM) enables nanoscale 3D imaging of biological structures but remains constrained by acquisition trade-offs, leading to anisotropic volumes with limited axial resolution. Existing deep learning methods seek to restore isotropy by leveraging lateral priors, yet their assumptions break down for morphologically anisotropic structures. We present EMGauss, a general framework for 3D reconstruction from planar scanned 2D slices with applications in vEM, which circumvents the inherent limitations of isotropy-based approaches. Our key innovation is to reframe slice-to-3D reconstruction as a 3D dynamic scene rendering problem based on Gaussian splatting, where the progression of axial slices is modeled as the temporal evolution of 2D Gaussian point clouds. To enhance fidelity in data-sparse regimes, we incorporate a Teacher-Student bootstrapping mechanism that uses high-confidence predictions on unobserved slices as pseudo-supervisory signals. Compared with diffusion- and GAN-based reconstruction methods, EMGauss substantially improves interpolation quality, enables continuous slice synthesis, and eliminates the need for large-scale pretraining. Beyond vEM, it potentially provides a generalizable slice-to-3D solution across diverse imaging domains. --- ### Mechanistic Interpretability of GPT-2: Lexical and Contextual Layers in Sentiment Analysis Authors: Amartya Hatua Published: 2025-12-07 Categories: cs.CL, cs.AI arXiv: https://arxiv.org/abs/2512.06681v1 We present a mechanistic interpretability study of GPT-2 that causally examines how sentiment information is processed across its transformer layers. Using systematic activation patching across all 12 layers, we test the hypothesized two-stage sentiment architecture comprising early lexical detection and mid-layer contextual integration. Our experiments confirm that early layers (0-3) act as lexical sentiment detectors, encoding stable, position specific polarity signals that are largely independent of context. However, all three contextual integration hypotheses: Middle Layer Concentration, Phenomenon Specificity, and Distributed Processing are falsified. Instead of mid-layer specialization, we find that contextual phenomena such as negation, sarcasm, domain shifts etc. are integrated primarily in late layers (8-11) through a unified, non-modular mechanism. These experimental findings provide causal evidence that GPT-2's sentiment computation differs from the predicted hierarchical pattern, highlighting the need for further empirical characterization of contextual integration in large language models. --- ### CMV-Fuse: Cross Modal-View Fusion of AMR, Syntax, and Knowledge Representations for Aspect Based Sentiment Analysis Authors: Smitha Muthya Sudheendra, Mani Deep Cherukuri, Jaideep Srivastava Published: 2025-12-07 Categories: cs.CL arXiv: https://arxiv.org/abs/2512.06679v2 Natural language understanding inherently depends on integrating multiple complementary perspectives spanning from surface syntax to deep semantics and world knowledge. However, current Aspect-Based Sentiment Analysis (ABSA) systems typically exploit isolated linguistic views, thereby overlooking the intricate interplay between structural representations that humans naturally leverage. We propose CMV-Fuse, a Cross-Modal View fusion framework that emulates human language processing by systematically combining multiple linguistic perspectives. Our approach systematically orchestrates four linguistic perspectives: Abstract Meaning Representations, constituency parsing, dependency syntax, and semantic attention, enhanced with external knowledge integration. Through hierarchical gated attention fusion across local syntactic, intermediate semantic, and global knowledge levels, CMV-Fuse captures both fine-grained structural patterns and broad contextual understanding. A novel structure aware multi-view contrastive learning mechanism ensures consistency across complementary representations while maintaining computational efficiency. Extensive experiments demonstrate substantial improvements over strong baselines on standard benchmarks, with analysis revealing how each linguistic view contributes to more robust sentiment analysis. --- ### GradientSpace: Unsupervised Data Clustering for Improved Instruction Tuning Authors: Shrihari Sridharan, Deepak Ravikumar, Anand Raghunathan, Kaushik Roy Published: 2025-12-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.06678v1 Instruction tuning is one of the key steps required for adapting large language models (LLMs) to a broad spectrum of downstream applications. However, this procedure is difficult because real-world datasets are rarely homogeneous; they consist of a mixture of diverse information, causing gradient interference, where conflicting gradients pull the model in opposing directions, degrading performance. A common strategy to mitigate this issue is to group data based on semantic or embedding similarity. However, this fails to capture how data influences model parameters during learning. While recent works have attempted to cluster gradients directly, they randomly project gradients into lower dimensions to manage memory, which leads to accuracy loss. Moreover, these methods rely on expert ensembles which necessitates multiple inference passes and expensive on-the-fly gradient computations during inference. To address these limitations, we propose GradientSpace, a framework that clusters samples directly in full-dimensional gradient space. We introduce an online SVD-based algorithm that operates on LoRA gradients to identify latent skills without the infeasible cost of storing all sample gradients. Each cluster is used to train a specialized LoRA expert along with a lightweight router trained to select the best expert during inference. We show that routing to a single, appropriate expert outperforms expert ensembles used in prior work, while significantly reducing inference latency. Our experiments across mathematical reasoning, code generation, finance, and creative writing tasks demonstrate that GradientSpace leads to coherent expert specialization and consistent accuracy gains over state-of-the-art clustering methods and finetuning techniques. --- ### FedDSR: Federated Deep Supervision and Regularization Towards Autonomous Driving Authors: Wei-Bin Kou, Guangxu Zhu, Bingyang Cheng, Chen Zhang, Yik-Chung Wu, Jianping Wang Published: 2025-12-07 Categories: cs.RO, cs.LG arXiv: https://arxiv.org/abs/2512.06676v1 Federated Learning (FL) enables collaborative training of autonomous driving (AD) models across distributed vehicles while preserving data privacy. However, FL encounters critical challenges such as poor generalization and slow convergence due to non-independent and identically distributed (non-IID) data from diverse driving environments. To overcome these obstacles, we introduce Federated Deep Supervision and Regularization (FedDSR), a paradigm that incorporates multi-access intermediate layer supervision and regularization within federated AD system. Specifically, FedDSR comprises following integral strategies: (I) to select multiple intermediate layers based on predefined architecture-agnostic standards. (II) to compute mutual information (MI) and negative entropy (NE) on those selected layers to serve as intermediate loss and regularizer. These terms are integrated into the output-layer loss to form a unified optimization objective, enabling comprehensive optimization across the network hierarchy. (III) to aggregate models from vehicles trained based on aforementioned rules of (I) and (II) to generate the global model on central server. By guiding and penalizing the learning of feature representations at intermediate stages, FedDSR enhances the model generalization and accelerates model convergence for federated AD. We then take the semantic segmentation task as an example to assess FedDSR and apply FedDSR to multiple model architectures and FL algorithms. Extensive experiments demonstrate that FedDSR achieves up to 8.93% improvement in mIoU and 28.57% reduction in training rounds, compared to other FL baselines, making it highly suitable for practical deployment in federated AD ecosystems. --- ### RunawayEvil: Jailbreaking the Image-to-Video Generative Models Authors: Songping Wang, Rufan Qian, Yueming Lyu, Qinglong Liu, Linzhuang Zou, Jie Qin, Songhua Liu, Caifeng Shan Published: 2025-12-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.06674v1 Image-to-Video (I2V) generation synthesizes dynamic visual content from image and text inputs, providing significant creative control. However, the security of such multimodal systems, particularly their vulnerability to jailbreak attacks, remains critically underexplored. To bridge this gap, we propose RunawayEvil, the first multimodal jailbreak framework for I2V models with dynamic evolutionary capability. Built on a "Strategy-Tactic-Action" paradigm, our framework exhibits self-amplifying attack through three core components: (1) Strategy-Aware Command Unit that enables the attack to self-evolve its strategies through reinforcement learning-driven strategy customization and LLM-based strategy exploration; (2) Multimodal Tactical Planning Unit that generates coordinated text jailbreak instructions and image tampering guidelines based on the selected strategies; (3) Tactical Action Unit that executes and evaluates the multimodal coordinated attacks. This self-evolving architecture allows the framework to continuously adapt and intensify its attack strategies without human intervention. Extensive experiments demonstrate RunawayEvil achieves state-of-the-art attack success rates on commercial I2V models, such as Open-Sora 2.0 and CogVideoX. Specifically, RunawayEvil outperforms existing methods by 58.5 to 79 percent on COCO2017. This work provides a critical tool for vulnerability analysis of I2V models, thereby laying a foundation for more robust video generation systems. --- ### Detector-Empowered Video Large Language Model for Efficient Spatio-Temporal Grounding Authors: Shida Gao, Feng Xue, Xiangfeng Wang, Anlong Ming, Zhaowen Lin, Haiyang Zhang, Teng Long, Nicu Sebe, Yihua Shao, Haozhe Wang, Wei Wang Published: 2025-12-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.06673v2 Multimodal large language models (MLLMs) are rapidly expanding from general video understanding to finer-grained understanding such as spatio-temporal video grounding (STVG) and reasoning. In these tasks, an MLLM must localize the user-queried target in time and space and take the results as evidence for reasoning. Existing MLLM methods mainly follow two paradigms: (1) Direct Localization, which outputs STVG results with extra alignment modules or specialized decoders; and (2) Candidate-based Selection, which first constructs tube-level candidates and then selects the relevant one by an MLLM. However, both suffer from a serious efficiency bottleneck: the former incurs linearly growing decoding cost as the queried temporal span increases, while the latter relies on costly candidate construction. To break this bottleneck, we propose DEViL, a detector-empowered Video-LLM with a simple key idea: offloading dense spatial grounding from the MLLM to a fully parallelizable, well-trained detector. Specifically, DEViL distills the query into a detector-compatible reference-semantic token, which replaces the detector's text embedding to enable spatial grounding in a single pass. Then, we design temporal consistency regularization to match objects across frames and enforce their coherence over time. In this way, DEViL avoids long coordinate decoding and heavy candidate pipelines. Extensive experiments show that DEViL achieves strong performance (43.1% m_vIoU on HC-STVG) with superior efficiency (14.33 FPS), while preserving the general reasoning capacity of the MLLM backbone. --- ### The Meta-Learning Gap: Combining Hydra and Quant for Large-Scale Time Series Classification Authors: Urav Maniar Published: 2025-12-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.06666v1 Time series classification faces a fundamental trade-off between accuracy and computational efficiency. While comprehensive ensembles like HIVE-COTE 2.0 achieve state-of-the-art accuracy, their 340-hour training time on the UCR benchmark renders them impractical for large-scale datasets. We investigate whether targeted combinations of two efficient algorithms from complementary paradigms can capture ensemble benefits while maintaining computational feasibility. Combining Hydra (competing convolutional kernels) and Quant (hierarchical interval quantiles) across six ensemble configurations, we evaluate performance on 10 large-scale MONSTER datasets (7,898 to 1,168,774 training instances). Our strongest configuration improves mean accuracy from 0.829 to 0.836, succeeding on 7 of 10 datasets. However, prediction-combination ensembles capture only 11% of theoretical oracle potential, revealing a substantial meta-learning optimization gap. Feature-concatenation approaches exceeded oracle bounds by learning novel decision boundaries, while prediction-level complementarity shows moderate correlation with ensemble gains. The central finding: the challenge has shifted from ensuring algorithms are different to learning how to combine them effectively. Current meta-learning strategies struggle to exploit the complementarity that oracle analysis confirms exists. Improved combination strategies could potentially double or triple ensemble gains across diverse time series classification applications. --- ### Rethinking Robustness: A New Approach to Evaluating Feature Attribution Methods Authors: Panagiota Kiourti, Anu Singh, Preeti Duraipandian, Weichao Zhou, Wenchao Li Published: 2025-12-07 Categories: cs.LG, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2512.06665v1 This paper studies the robustness of feature attribution methods for deep neural networks. It challenges the current notion of attributional robustness that largely ignores the difference in the model's outputs and introduces a new way of evaluating the robustness of attribution methods. Specifically, we propose a new definition of similar inputs, a new robustness metric, and a novel method based on generative adversarial networks to generate these inputs. In addition, we present a comprehensive evaluation with existing metrics and state-of-the-art attribution methods. Our findings highlight the need for a more objective metric that reveals the weaknesses of an attribution method rather than that of the neural network, thus providing a more accurate evaluation of the robustness of attribution methods. --- ### CoT4Det: A Chain-of-Thought Framework for Perception-Oriented Vision-Language Tasks Authors: Yu Qi, Yumeng Zhang, Chenting Gong, Xiao Tan, Weiming Zhang, Wei Zhang, Jingdong Wang Published: 2025-12-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.06663v1 Large Vision-Language Models (LVLMs) have demonstrated remarkable success in a broad range of vision-language tasks, such as general visual question answering and optical character recognition (OCR). However, their performance on perception-centric tasks -- such as object detection, semantic segmentation, and depth estimation -- remains significantly inferior to that of task-specific expert models. For example, Qwen2.5-VL-7B-Instruct achieves only 19% mAP on COCO2017 val, particularly struggling with dense scenes and small object recall. In this work, we introduce Chain-of-Thought for Detection (CoT4Det), a simple but efficient strategy that reformulates perception tasks into three interpretable steps: classification, counting, and grounding -- each more naturally aligned with the reasoning capabilities of LVLMs. Extensive experiments demonstrate that our method significantly improves perception performance without compromising general vision language capabilities. With a standard Qwen2.5-VL-7B-Instruct, CoT4Det boosts mAP from 19.0% to 33.0% on COCO2017 val and achieves competitive results across a variety of perception benchmarks, outperforming baselines by +2% on RefCOCO series and 19% on Flickr30k entities. --- ### Personalized Image Descriptions from Attention Sequences Authors: Ruoyu Xue, Hieu Le, Jingyi Xu, Sounak Mondal, Abe Leite, Gregory Zelinsky, Minh Hoai, Dimitris Samaras Published: 2025-12-07 Categories: cs.CV arXiv: https://arxiv.org/abs/2512.06662v1 People can view the same image differently: they focus on different regions, objects, and details in varying orders and describe them in distinct linguistic styles. This leads to substantial variability in image descriptions. However, existing models for personalized image description focus on linguistic style alone, with no prior work leveraging individual viewing patterns. We address this gap by explicitly modeling personalized viewing behavior as a core factor in description generation. Our method, DEPER (DEscription-PERception persona encoder), learns a subject embedding that captures both linguistic style and viewing behavior, guided by an auxiliary attention-prediction task. A lightweight adapter aligns these embeddings with a frozen vision-language model, enabling few-shot personalization without retraining. Across four datasets spanning diverse viewing tasks and both short and detailed descriptions, DEPER achieves a 24% average improvement, showing that modeling personalized attention produces more human-aligned and high-quality descriptions. We posit that understanding how people see helps predict what they say; modeling human diversity in perception can improve both performance and human alignment in multimodal systems. --- ### Effective and Efficient Threat Hunting with Small Language Models Authors: Saleha Muzammil, Rahul Reddy, Vishal Kamalakrishnan, Hadi Ahmadi, Wajih Ul Hassan Published: 2025-12-07 Categories: cs.CR, cs.AI arXiv: https://arxiv.org/abs/2512.06660v3 Analysts in Security Operations Centers query massive telemetry streams using Kusto Query Language (KQL), but writing correct KQL demands specialized expertise that bottlenecks scaling security teams. We investigate how Small Language Models (SLMs) can enable accurate, cost-effective translation from natural language queries (NLQs) to KQL. We propose a three-knob framework spanning prompting, fine-tuning, and architecture. First, we adapt NL2KQL for SLMs with lightweight retrieval and introduce error-aware prompting that targets common parser failures with a handful of mined tips, at a fraction of the tokens KQL's full rule set would require. Second, we apply LoRA fine-tuning with rationale distillation augmenting each NLQ-KQL pair with a brief chain-of-thought to transfer teacher reasoning. This yields an informative negative result, as neither variant surpasses targeted prompting. Third, we propose a two-stage architecture pairing an SLM drafter with a low-cost LLM judge for schema-aware refinement. We evaluate nine models (five SLMs, four LLMs) on syntax correctness, semantic accuracy, table selection, filter precision, latency, and token cost. On Microsoft's NL2KQL Defender Evaluation dataset, our two-stage approach reaches 0.987 syntax and 0.906 schema-valid ("semantic") accuracy, exceeding every baseline we run under equivalent infrastructure, and it generalizes to independently authored queries over the same schema (0.964 syntax, 0.831 schema-valid). The only baselines within 0.05 schema-valid are NL2KQL+GPT-4o (0.878) and NL2KQL+GPT-5 (0.861), which cost USD 2.998 and USD 2.018 for 230 queries against USD 0.213 for ours, a 9.5-14x reduction at matched accuracy. These results establish SLMs as a practical foundation for natural-language querying in security operations. --- ### Phase-multiplexed optical computing: Reconfiguring a multi-task diffractive optical processor using illumination phase diversity Authors: Xiao Wang, Aydogan Ozcan Published: 2025-12-07 Categories: physics.optics, cs.NE, physics.app-ph arXiv: https://arxiv.org/abs/2512.06658v1 We report a monochrome multi-task diffractive network architecture that leverages illumination phase multiplexing to dynamically reconfigure its output function and accurately implement a large group of complex-valued linear transformations between an input and output aperture. Each member of the desired group of T unique transformations is encoded and addressed with a distinct 2D illumination phase profile, termed "phase key", which illuminates the input aperture, activating the corresponding transformation at the output field-of-view. A common diffractive optical network, optimized with T phase keys, demultiplexes these encoded inputs and accurately executes any of the T distinct linear transformations at its output. We demonstrate that a diffractive network composed of N = 2 x T x Ni x No optimized diffractive features can realize T distinct complex-valued linear transformations, accurately executed for any complex field at the input aperture, where Ni and No refer to the input/output pixels, respectively. In our proof-of-concept numerical analysis, T = 512 complex-valued transformations are implemented by the same monochrome diffractive network with negligible error using illumination phase diversity. Compared with wavelength-multiplexed diffractive systems, phase-multiplexing architecture significantly lowers the transformation errors, potentially enabling larger-scale optical transformations to be implemented through a monochrome processor. Phase-multiplexed multi-task diffractive networks would enhance the capabilities of optical computing and machine-vision systems. --- ### TextMamba: Scene Text Detector with Mamba Authors: Qiyan Zhao, Yue Yan, Da-Han Wang Published: 2025-12-07 Categories: cs.CV, cs.AI arXiv: https://arxiv.org/abs/2512.06657v1 In scene text detection, Transformer-based methods have addressed the global feature extraction limitations inherent in traditional convolution neural network-based methods. However, most directly rely on native Transformer attention layers as encoders without evaluating their cross-domain limitations and inherent shortcomings: forgetting important information or focusing on irrelevant representations when modeling long-range dependencies for text detection. The recently proposed state space model Mamba has demonstrated better long-range dependencies modeling through a linear complexity selection mechanism. Therefore, we propose a novel scene text detector based on Mamba that integrates the selection mechanism with attention layers, enhancing the encoder's ability to extract relevant information from long sequences. We adopt the Top\_k algorithm to explicitly select key information and reduce the interference of irrelevant information in Mamba modeling. Additionally, we design a dual-scale feed-forward network and an embedding pyramid enhancement module to facilitate high-dimensional hidden state interactions and multi-scale feature fusion. Our method achieves state-of-the-art or competitive performance on various benchmarks, with F-measures of 89.7\%, 89.2\%, and 78.5\% on CTW1500, TotalText, and ICDAR19ArT, respectively. Codes will be available. --- ### The Online Discourse of Virtual Reality and Anxiety Authors: Kwabena Yamoah, Cass Dykeman Published: 2025-12-07 Categories: cs.CL, stat.CO arXiv: https://arxiv.org/abs/2512.06656v1 VR in the treatment of clinical concerns such as generalized anxiety disorder or social anxiety. VR has created additional pathways to support patient well-being and care. Understanding online discussion of what users think about this technology may further support its efficacy. The purpose of this study was to employ a corpus linguistic methodology to identify the words and word networks that shed light on the online discussion of virtual reality and anxiety. Using corpus linguistics, frequently used words in discussion along with collocation were identified by utilizing Sketch Engine software. The results of the study, based upon the English Trends corpus, identified VR, Oculus, and headset as the most frequently discussed within the VR and anxiety subcorpus. These results point to the development of the virtual system, along with the physical apparatus that makes viewing and engaging with the virtual environment possible. Additional results point to collocation of prepositional phrases such as of virtual reality, in virtual reality, and for virtual reality relating to the design, experience, and development, respectively. These findings offer new perspective on how VR and anxiety together are discussed in general discourse and offer pathways for future opportunities to support counseling needs through development and accessibility. Keywords: anxiety disorders, corpus linguistics, Sketch Engine, and virtual reality VR --- ### Graph-Regularized Sparse Autoencoders for LLM Safety Steering Authors: Jehyeok Yeon, Federico Cinus, Yifan Wu, Luca Luceri Published: 2025-12-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.06655v3 Sparse autoencoders (SAEs) are increasingly used to extract activation directions for inference-time steering, but their standard sparsity objective treats latent features as independent. This prior can be poorly matched to high-level safety behaviors, where refusal and harmful compliance appear to depend on distributed structure in activation space. We introduce Graph-Regularized Sparse Autoencoders (GSAE), a dictionary-learning method that learns safety-steering directions by smoothing SAE decoder vectors over a neuron co-activation graph and applying the resulting direction bank through a two-gate runtime controller. Empirically, GSAE improves selective refusal across JailbreakBench, HarmBench, and XSTest, increasing harmful-request refusal while keeping benign-prompt refusals low. On Llama-3-8B, replacing the standard SAE with GSAE in an otherwise identical pipeline improves $Δ_s$ by $20.1$ points on JailbreakBench and $16.8$ points on HarmBench. GSAE outperforms activation-steering baselines and black-box guardrails, preserves benign-task performance, generalizes across Llama-3, Mistral, Qwen 2.5, and Phi-4, and remains strong under black-box and gray-box jailbreak attacks. --- ### LightSearcher: Efficient DeepSearch via Experiential Memory Authors: Hengzhi Lan, Yue Yu, Li Qian, Li Peng, Jie Wu, Wei Liu, Jian Luan, Ting Bai Published: 2025-12-07 Categories: cs.AI arXiv: https://arxiv.org/abs/2512.06653v3 DeepSearch paradigms have become a core enabler for deep reasoning models, allowing them to invoke external search tools to access up-to-date, domain-specific knowledge beyond parametric boundaries, thereby enhancing the depth and factual reliability of reasoning. Building upon this foundation, recent advances in reinforcement learning (RL) have further empowered models to autonomously and strategically control search tool usage, optimizing when and how to query external knowledge sources. Yet, these RL-driven DeepSearch systems often reveal a see-saw trade-off between accuracy and efficiency-frequent tool invocations can improve factual correctness but lead to unnecessary computational overhead and diminished efficiency. To address this challenge, we propose LightSearcher, an efficient RL framework that incorporates textual experiential memory by learning contrastive reasoning trajectories to generate interpretable summaries of successful reasoning patterns. In addition, it employs an adaptive reward shaping mechanism that penalizes redundant tool calls only in correct-answer scenarios. This design effectively balances the inherent accuracy-efficiency trade-off in DeepSearch paradigms. Experiments on four multi-hop QA benchmarks show that LightSearcher maintains accuracy comparable to SOTA baseline ReSearch, while reducing search tool invocations by 39.6%, inference time by 48.6%, and token consumption by 21.2%, demonstrating its superior efficiency. --- ### Adaptive Test-Time Training for Predicting Need for Invasive Mechanical Ventilation in Multi-Center Cohorts Authors: Xiaolei Lu, Shamim Nemati Published: 2025-12-07 Categories: cs.LG, cs.AI arXiv: https://arxiv.org/abs/2512.06652v2 Accurate prediction of the need for invasive mechanical ventilation (IMV) in intensive care units (ICUs) patients is crucial for timely interventions and resource allocation. However, variability in patient populations, clinical practices, and electronic health record (EHR) systems across institutions introduces domain shifts that degrade the generalization performance of predictive models during deployment. Test-Time Training (TTT) has emerged as a promising approach to mitigate such shifts by adapting models dynamically during inference without requiring labeled target-domain data. In this work, we introduce Adaptive Test-Time Training (AdaTTT), an enhanced TTT framework tailored for EHR-based IMV prediction in ICU settings. We begin by deriving information-theoretic bounds on the test-time prediction error and demonstrate that it is constrained by the uncertainty between the main and auxiliary tasks. To enhance their alignment, we introduce a self-supervised learning framework with pretext tasks: reconstruction and masked feature modeling optimized through a dynamic masking strategy that emphasizes features critical to the main task. Additionally, to improve robustness against domain shifts, we incorporate prototype learning and employ Partial Optimal Transport (POT) for flexible, partial feature alignment while maintaining clinically meaningful patient representations. Experiments across multi-center ICU cohorts demonstrate competitive classification performance on different test-time adaptation benchmarks. --- ### Estimating Black Carbon Concentration from Urban Traffic Using Vision-Based Machine Learning Authors: Camellia Zakaria, Aryan Sadeghi, Weaam Jaafar, Junshi Xu, Alex Mariakakis, Marianne Hatzopoulou Published: 2025-12-07 Categories: cs.LG, cs.CV, cs.CY, cs.ET arXiv: https://arxiv.org/abs/2512.06649v1 Black carbon (BC) emissions in urban areas are primarily driven by traffic, with hotspots near major roads disproportionately affecting marginalized communities. Because BC monitoring is typically performed using costly and specialized instruments. there is little to no available data on BC from local traffic sources that could help inform policy interventions targeting local factors. By contrast, traffic monitoring systems are widely deployed in cities around the world, highlighting the imbalance between what we know about traffic conditions and what do not know about their environmental consequences. To bridge this gap, we propose a machine learning-driven system that extracts visual information from traffic video to capture vehicles behaviors and conditions. Combining these features with weather data, our model estimates BC at street level, achieving an R-squared value of 0.72 and RMSE of 129.42 ng/m3 (nanogram per cubic meter). From a sustainability perspective, this work leverages resources already supported by urban infrastructure and established modeling techniques to generate information relevant to traffic emission. Obtaining BC concentration data provides actionable insights to support pollution reduction, urban planning, public health, and environmental justice at the local municipal level. --- ### Financial Fraud Identification and Interpretability Study for Listed Companies Based on Convolutional Neural Network Authors: Xiao Li Published: 2025-12-07 Categories: cs.LG, cs.AI, cs.CV arXiv: https://arxiv.org/abs/2512.06648v2 Since the emergence of joint-stock companies, financial fraud by listed firms has repeatedly undermined capital markets. Fraud is difficult to detect because of covert tactics and the high labor and time costs of audits. Traditional statistical models are interpretable but struggle with nonlinear feature interactions, while machine learning models are powerful but often opaque. In addition, most existing methods judge fraud only for the current year based on current year data, limiting timeliness. This paper proposes a financial fraud detection framework for Chinese A-share listed companies based on convolutional neural networks (CNNs). We design a feature engineering scheme that transforms firm-year panel data into image like representations, enabling the CNN to capture cross-sectional and temporal patterns and to predict fraud in advance. Experiments show that the CNN outperforms logistic regression and LightGBM in accuracy, robustness, and early-warning performance, and that proper tuning of the classification threshold is crucial in high-risk settings. To address interpretability, we analyze the model along the dimensions of entity, feature, and time using local explanation techniques. We find that solvency, ratio structure, governance structure, and internal control are general predictors of fraud, while environmental indicators matter mainly in high-pollution industries. Non-fraud firms share stable feature patterns, whereas fraud firms exhibit heterogeneous patterns concentrated in short time windows. A case study of Guanong Shares in 2022 shows that cash flow analysis, social responsibility, governance structure, and per-share indicators are the main drivers of the model's fraud prediction, consistent with the company's documented misconduct. --- ### Masked Autoencoder Pretraining on Strong-Lensing Images for Joint Dark-Matter Model Classification and Super-Resolution Authors: Achmad Ardani Prasha, Clavino Ourizqi Rachmadi, Muhamad Fauzan Ibnu Syahlan, Naufal Rahfi Anugerah, Nanda Garin Raditya, Putri Amelia, Sabrina Laila Mutiara, Hilman Syachr Ramadhan Published: 2025-12-07 Categories: cs.CV, astro-ph.CO, astro-ph.IM, cs.AI, cs.LG arXiv: https://arxiv.org/abs/2512.06642v1 Strong gravitational lensing can reveal the influence of dark-matter substructure in galaxies, but analyzing these effects from noisy, low-resolution images poses a significant challenge. In this work, we propose a masked autoencoder (MAE) pretraining strategy on simulated strong-lensing images from the DeepLense ML4SCI benchmark to learn generalizable representations for two downstream tasks: (i) classifying the underlying dark matter model (cold dark matter, axion-like, or no substructure) and (ii) enhancing low-resolution lensed images via super-resolution. We pretrain a Vision Transformer encoder using a masked image modeling objective, then fine-tune the encoder separately for each task. Our results show that MAE pretraining, when combined with appropriate mask ratio tuning, yields a shared encoder that matches or exceeds a ViT trained from scratch. Specifically, at a 90% mask ratio, the fine-tuned classifier achieves macro AUC of 0.968 and accuracy of 88.65%, compared to the scratch baseline (AUC 0.957, accuracy 82.46%). For super-resolution (16x16 to 64x64), the MAE-pretrained model reconstructs images with PSNR ~33 dB and SSIM 0.961, modestly improving over scratch training. We ablate the MAE mask ratio, revealing a consistent trade-off: higher mask ratios improve classification but slightly degrade reconstruction fidelity. Our findings demonstrate that MAE pretraining on physics-rich simulations provides a flexible, reusable encoder for multiple strong-lensing analysis tasks. --- ### An Index-based Approach for Efficient and Effective Web Content Extraction Authors: Yihan Chen, Benfeng Xu, Xiaorui Wang, Zhendong Mao Published: 2025-12-07 Categories: cs.IR, cs.CL arXiv: https://arxiv.org/abs/2512.06641v1 As web agents (e.g., Deep Research) routinely consume massive volumes of web pages to gather and analyze information, LLM context management -- under large token budgets and low signal density -- emerges as a foundational, high-importance, and technically challenging problem for agentic and RAG pipelines. Existing solutions for extracting relevant content are inadequate: generative extraction models suffer from high latency, rule-based heuristics lack adaptability, and chunk-and-rerank methods are blind to webpage structure. To overcome these issues, we introduce Index-based Web Content Extraction to reframe the extraction process from slow, token-by-token generation into a highly efficient, discriminative task of index prediction, achieving both effectiveness and efficiency. We partition HTML into structure-aware, addressable segments, and extract only the positional indices of content relevant to a given query. This method decouples extraction latency from content length, enabling rapid, query-relevant extraction. We first evaluate our method as a post-retrieval processing component within an RAG QA system and find that it improves QA accuracy. Then we directly measure its match rate with the target content in two scenarios: main content extraction (ME) and query-relevant extraction (QE). Experimental results show that our method outperforms existing works in both accuracy and speed, effectively bridging the gap between LLMs and the vast webpages. --- ### Learning to Hedge Swaptions Authors: Zaniar Ahmadi, Frédéric Godin Published: 2025-12-07 Categories: q-fin.RM, cs.LG arXiv: https://arxiv.org/abs/2512.06639v1 This paper investigates the deep hedging framework, based on reinforcement learning (RL), for the dynamic hedging of swaptions, contrasting its performance with traditional sensitivity-based rho-hedging. We design agents under three distinct objective functions (mean squared error, downside risk, and Conditional Value-at-Risk) to capture alternative risk preferences and evaluate how these objectives shape hedging styles. Relying on a three-factor arbitrage-free dynamic Nelson-Siegel model for our simulation experiments, our findings show that near-optimal hedging effectiveness is achieved when using two swaps as hedging instruments. Deep hedging strategies dynamically adapt the hedging portfolio's exposure to risk factors across states of the market. In our experiments, their out-performance over rho-hedging strategies persists even in the presence some of model misspecification. These results highlight RL's potential to deliver more efficient and resilient swaption hedging strategies. --- ### The Impact of Data Characteristics on GNN Evaluation for Detecting Fake News Authors: Isha Karn, David Jensen Published: 2025-12-07 Categories: cs.LG arXiv: https://arxiv.org/abs/2512.06638v1 Graph neural networks (GNNs) are widely used for the detection of fake news by modeling the content and propagation structure of news articles on social media. We show that two of the most commonly used benchmark data sets - GossipCop and PolitiFact - are poorly suited to evaluating the utility of models that use propagation structure. Specifically, these data sets exhibit shallow, ego-like graph topologies that provide little or no ability to differentiate among modeling methods. We systematically benchmark five GNN architectures against a structure-agnostic multilayer perceptron (MLP) that uses the same node features. We show that MLPs match or closely trail the performance of GNNs, with performance gaps often within 1-2% and overlapping confidence intervals. To isolate the contribution of structure in these datasets, we conduct controlled experiments where node features are shuffled or edge structures randomized. We find that performance collapses under feature shuffling but remains stable under edge randomization. This suggests that structure plays a negligible role in these benchmarks. Structural analysis further reveals that over 75% of nodes are only one hop from the root, exhibiting minimal structural diversity. In contrast, on synthetic datasets where node features are noisy and structure is informative, GNNs significantly outperform MLPs. These findings provide strong evidence that widely used benchmarks do not meaningfully test the utility of modeling structural features, and they motivate the development of datasets with richer, more diverse graph topologies. --- ### Quantum Temporal Convolutional Neural Networks for Cross-Sectional Equity Return Prediction: A Comparative Benchmark Study Authors: Chi-Sheng Chen, Xinyu Zhang, En-Jui Kuo, Rong Fu, Qiuzhe Xie, Fan Zhang Published: 2025-12-07 Categories: cs.LG, quant-ph arXiv: https://arxiv.org/abs/2512.06630v2 Quantum machine learning offers a promising pathway for enhancing stock market prediction, particularly under complex, noisy, and highly dynamic financial environments. However, many classical forecasting models struggle with noisy input, regime shifts, and limited generalization capacity. To address these challenges, we propose a Quantum Temporal Convolutional Neural Network (QTCNN) that combines a classical temporal encoder with parameter-efficient quantum convolution circuits for cross-sectional equity return prediction. The temporal encoder extracts multi-scale patterns from sequential technical indicators, while the quantum processing leverages superposition and entanglement to enhance feature representation and suppress overfitting. We conduct a comprehensive benchmarking study on the JPX Tokyo Stock Exchange dataset and evaluate predictions through long-short portfolio construction using out-of-sample Sharpe ratio as the primary performance metric. QTCNN achieves a Sharpe ratio of 0.538, outperforming the best classical baseline by approximately 72\%. These results highlight the practical potential of quantum-enhanced forecasting model, QTCNN, for robust decision-making in quantitative finance. --- ### FlatFormer: A Flat Transformer Knowledge Tracing Model Based on Cognitive Bias Injection Authors: Xiao-li Xia, Hou-biao Li Published: 2025-12-07 Categories: cs.AI, cs.IT arXiv: https://arxiv.org/abs/2512.06629v1 Knowledge Tracing (KT) models face a critical ``Performance-Complexity Trap'': capturing complex cognitive dynamics like learning sessions and memory decay typically requires deep hierarchical architectures, which incur prohibitive computational costs for real-time deployment. To resolve this, we propose FlatFormer, a streamlined architecture based on the novel design paradigm of ``Information Injection over Structural Stacking.'' Unlike parameter-heavy hierarchical models, FlatFormer leverages a standard flat Transformer augmented with two lightweight injection mechanisms: (i) a hybrid input encoding strategy combining learnable session identifiers with fixed sinusoidal step embeddings; and (ii) a pre-computed power-law bias integrated directly into attention logits to explicitly model the forgetting curve. Extensive experiments on four large-scale datasets (e.g., EdNet, Junyi) show that FlatFormer achieves state-of-the-art performance. For example, on the EdNet dataset, compared to the strongest hierarchical baseline (HiTSKT), its absolute AUC increased by 8.3%, while using less than 15% of parameters, and inference speed was about three times faster. These results validate that high cognitive fidelity does not necessitate architectural complexity. --- ### MIND-V: Hierarchical World Model for Long-Horizon Robotic Manipulation with RL-based Physical Alignment Authors: Ruicheng Zhang, Mingyang Zhang, Jun Zhou, Xiaofan Liu, Zunnan Xu, Zhizhou Zhong, Puxin Yan, Haocheng Luo, Xiu Li Published: 2025-12-07 Categories: cs.RO, cs.CV arXiv: https://arxiv.org/abs/2512.06628v4 Scalable embodied intelligence is constrained by the scarcity of diverse, long-horizon robotic manipulation data. Existing video world models in this domain are limited to synthesizing short clips of simple actions and often rely on manually defined trajectories. To this end, we introduce MIND-V, a cognitive hierarchical world model designed to synthesize physically plausible and logically coherent videos of long-horizon robotic manipulation. Inspired by cognitive science, MIND-V bridges high-level reasoning with pixel-level synthesis through three core components: a Semantic Reasoning Hub (SRH) that leverages a pre-trained vision-language model for task planning; a Behavioral Semantic Bridge (BSB) that translates abstract instructions into domain-invariant representations; and a Motor Video Generator (MVG) for conditional video rendering. MIND-V employs Staged Visual Future Rollouts, a test-time optimization strategy to enhance long-horizon robustness. To enforce adherence to physical laws, we introduce a GRPO reinforcement learning post-training phase guided by a novel Physical Foresight Coherence (PFC) reward. PFC leverages the V-JEPA2 world model as a physics referee to penalize implausible dynamics in the latent feature space. Experiments confirm MIND-V's SOTA performance in long-horizon simulation and its significant value for policy learning, introducing a scalable and fully autonomous framework for embodied data synthesis. ---