Machine Learning · Cross-listed from cs.CV
Maijunxian Wang, Ruisi Wang, Juyi Lin et al. (+53 more)
Rapid progress in
video models has largely focused on visual quality, leaving their
reasoning capabilities underexplored.
Video reasoning grounds intelligence in spatiotemporally consistent visual environments that go beyond what text can naturally capture, enabling intuitive
reasoning over spatiotemporal structure such as continuity, interaction, and causality. However, systematically studying
video reasoning and its scaling behavior is hindered by the lack of large-scale training data. To address this gap, we introduce the Very Big
Video Reasoning (VBVR)
Dataset, an unprecedentedly large-scale resource spanning 200 curated
reasoning tasks following a principled taxonomy and over one million
video clips, approximately three orders of magnitude larger than existing
datasets. We further present VBVR-Bench, a verifiable evaluation framework that moves beyond model-based judging by incorporating rule-based, human-aligned scorers, enabling reproducible and interpretable diagnosis of
video reasoning capabilities. Leveraging the VBVR suite, we conduct one of the first large-scale scaling studies of
video reasoning and observe early signs of emergent generalization to unseen
reasoning tasks. Together, VBVR lays a foundation for the next stage of research in generalizable
video reasoning. The data,
benchmark toolkit, and models are publicly available at https://
video-reason.com/?v=vbvr .
Read the full abstract ▾
All authors Maijunxian Wang, Ruisi Wang, Juyi Lin, Ran Ji, Thaddäus Wiedemer, Qingying Gao, Dezhi Luo, Yaoyao Qian, Lianyu Huang, Zelong Hong, Jiahui Ge, Qianli Ma, Hang He, Yifan Zhou, Lingzi Guo, Lantao Mei, Jiachen Li, Hanwen Xing, Tianqi Zhao, Fengyuan Yu, Weihang Xiao, Yizheng Jiao, Jianheng Hou, Danyang Zhang, Pengcheng Xu, Boyang Zhong, Zehong Zhao, Gaoyun Fang, John Kitaoka, Yile Xu, Hua Xu, Kenton Blacutt, Tin Nguyen, Siyuan Song, Haoran Sun, Shaoyue Wen, Linyang He, Runming Wang, Yanzhi Wang, Mengyue Yang, Ziqiao Ma, Raphaël Millière, Freda Shi, Nuno Vasconcelos, Daniel Khashabi, Alan Yuille, Yilun Du, Ziming Liu, Bo Li, Dahua Lin, Ziwei Liu, Vikash Kumar, Yijiang Li, Lei Yang, Zhongang Cai, Hokin Deng
Machine Learning · Cross-listed from cs.CR
Skill-Inject: Measuring Agent Vulnerability to Skill File Attacks
David Schmotz, Luca Beurer-Kellner, Sahar Abdelnabi et al. (+1 more)
LLM
agents are evolving rapidly, powered by code execution, tools, and the recently introduced
agent skills feature. Skills allow users to extend LLM applications with specialized third-party code, knowledge, and instructions. Although this can extend
agent capabilities to new domains, it creates an increasingly complex
agent supply chain, offering new surfaces for
prompt injection attacks. We identify skill-based
prompt injection as a significant threat and introduce SkillInject, a
benchmark evaluating the susceptibility of widely-used LLM
agents to injections through skill files. SkillInject contains 202 injection-task pairs with attacks ranging from obviously malicious injections to subtle, context-dependent attacks hidden in otherwise legitimate instructions. We evaluate frontier LLMs on SkillInject, measuring both security in terms of harmful instruction avoidance and utility in terms of legitimate instruction compliance. Our results show that today's
agents are highly vulnerable with up to 80% attack success rate with frontier models, often executing extremely harmful instructions including data exfiltration, destructive action, and ransomware-like behavior. They furthermore suggest that this problem will not be solved through model scaling or simple input filtering, but that robust
agent security will require context-aware authorization frameworks. Our
benchmark is available at https://www.skill-inject.com/.
Read the full abstract ▾
All authors David Schmotz, Luca Beurer-Kellner, Sahar Abdelnabi, Maksym Andriushchenko
Machine Learning · Cross-listed from stat.ML
JUCAL: Jointly Calibrating Aleatoric and Epistemic Uncertainty in Classification Tasks
Jakob Heiss, Sören Lambrecht, Jakob Weissteiner et al. (+4 more)
We study post-calibration uncertainty for trained ensembles of classifiers. Specifically, we consider both aleatoric (label noise) and epistemic (model) uncertainty. Among the most popular and widely used calibration methods in classification are temperature scaling (i.e., pool-then-calibrate) and conformal methods. However, the main shortcoming of these calibration methods is that they do not balance the proportion of aleatoric and epistemic uncertainty. Not balancing these uncertainties can severely misrepresent predictive uncertainty, leading to overconfident predictions in some input regions while being underconfident in others. To address this shortcoming, we present a simple but powerful calibration
algorithm Joint Uncertainty Calibration (JUCAL) that jointly calibrates aleatoric and epistemic uncertainty. JUCAL jointly calibrates two constants to weight and scale epistemic and aleatoric uncertainties by optimizing the negative log-likelihood (NLL) on the validation/calibration
dataset. JUCAL can be applied to any trained ensemble of classifiers (e.g.,
transformers, CNNs, or tree-based methods), with minimal computational overhead, without requiring access to the models' internal parameters. We experimentally evaluate JUCAL on various text classification tasks, for ensembles of varying sizes and with different ensembling strategies. Our experiments show that JUCAL significantly outperforms SOTA calibration methods across all considered classification tasks, reducing NLL and predictive set size by up to 15% and 20%, respectively. Interestingly, even applying JUCAL to an ensemble of size 5 can outperform temperature-scaled ensembles of size up to 50 in terms of NLL and predictive set size, resulting in up to 10 times smaller inference costs. Thus, we propose JUCAL as a new go-to method for calibrating ensembles in classification.
Read the full abstract ▾
All authors Jakob Heiss, Sören Lambrecht, Jakob Weissteiner, Hanna Wutte, Žan Žurič, Josef Teichmann, Bin Yu
Machine Learning
Behavior Learning (BL): Learning Hierarchical Optimization Structures from Data
Zhenyao Ma, Yue Liang and Dongxu Li
Inspired by behavioral science, we propose Behavior Learning (BL), a novel general-purpose machine learning framework that learns interpretable and identifiable optimization structures from data, ranging from single optimization problems to hierarchical compositions. It unifies predictive performance, intrinsic interpretability, and identifiability, with broad applicability to scientific domains involving optimization. BL parameterizes a compositional utility function built from intrinsically interpretable modular blocks, which induces a data distribution for prediction and generation. Each block represents and can be written in symbolic form as a utility maximization problem (UMP), a foundational paradigm in behavioral science and a universal framework of optimization. BL supports architectures ranging from a single UMP to hierarchical compositions, the latter modeling hierarchical optimization structures. Its smooth and monotone variant (IBL) guarantees identifiability. Theoretically, we establish the universal approximation property of BL, and analyze the M-estimation properties of IBL. Empirically, BL demonstrates strong predictive performance, intrinsic interpretability and scalability to high-dimensional data. Code: https://github.com/MoonYLiang/Behavior-Learning ; install via pip install blnetwork.
Read the full abstract ▾
Machine Learning · Cross-listed from stat.ME
Conformal Risk Control for Non-Monotonic Losses
Anastasios N. Angelopoulos
Conformal risk control is an extension of conformal prediction for controlling risk functions beyond miscoverage. The original
algorithm controls the expected value of a loss that is monotonic in a one-dimensional parameter. Here, we present risk control guarantees for generic
algorithms applied to possibly non-monotonic losses with multidimensional parameters. The guarantees depend on the stability of the
algorithm – unstable
algorithms have looser guarantees. We give applications of this technique to selective image classification, FDR and IOU control of tumor
segmentations, and multigroup debiasing of recidivism predictions across overlapping race and sex groups using empirical risk minimization.
Read the full abstract ▾
Machine Learning
LAD: Learning Advantage Distribution for Reasoning
Wendi Li and Sharon Li
Current reinforcement learning objectives for large-model
reasoning primarily focus on maximizing expected rewards. This paradigm can lead to overfitting to dominant reward signals, while neglecting alternative yet valid
reasoning trajectories, thereby limiting diversity and exploration. To address this issue, we introduce Learning Advantage Distributions (LAD), a distribution-matching framework that replaces advantage maximization with learning the advantage-induced distribution. By establishing the equivalence between the optimal policy update and an advantage-based target distribution, we derive a practical LAD objective formulated as minimizing an $f$-divergence between the policy-induced and advantage-induced distributions. This yields a gradient update that increases likelihood for high-advantage responses while suppressing over-confident probability growth, preventing collapse without requiring auxiliary
entropy regularization. LAD incurs no extra training cost compared to GRPO and scales naturally to LLM post-training. In a controlled bandit setting, LAD faithfully recovers the
multimodal advantage distribution, validating the theoretical formulation. Experiments on math and code
reasoning tasks across several LLM backbones show that LAD reliably improves both accuracy and generative diversity.
Read the full abstract ▾
Machine Learning
Adaptation to Intrinsic Dependence in Diffusion Language Models
Yunxiao Zhao and Changxiao Cai
Diffusion language models (DLMs) have recently emerged as a promising alternative to autoregressive (AR) approaches, enabling parallel
token generation beyond a rigid left-to-right order. Despite growing empirical success, the theoretical understanding of how unmasking schedules – which specify the order and size of unmasked
tokens during sampling – affect generation quality remains limited. In this work, we introduce a distribution-agnostic unmasking schedule for DLMs that adapts to the (unknown) dependence structure of the target data distribution, without requiring any prior knowledge or hyperparameter tuning. In contrast to prior deterministic procedures that fix unmasking sizes, our method randomizes the number of
tokens revealed at each iteration. We show that, for two specific parameter choices, the sampling convergence guarantees – measured by Kullback-Leibler (KL) divergence – scale as $\widetilde O(\mathsf{TC}/K)$ and $\widetilde O(\mathsf{DTC}/K)$ respectively. Here, $K$ is the number of iterations, and $\mathsf{TC}$ and $\mathsf{DTC}$ are the total correlation and dual total correlation of the target distribution, capturing the intrinsic dependence structure underlying the data. Importantly, our guarantees hold in the practically relevant parallel-sampling regime $K<L$ where $L$ is the
token sequence length. These results significantly improve upon prior convergence theories and yield substantial sampling acceleration for low-complexity distributions. Overall, our findings unveil the adaptivity of DLMs to intrinsic data structures and shed light on the benefit of randomized unmasking sizes in inference schedule design.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.CL
NanoKnow: How to Know What Your Language Model Knows
Lingwei Gu, Nour Jedidi and Jimmy Lin
How do large
language models (LLMs) know what they know? Answering this question has been difficult because pre-training data is often a "black box" - unknown or inaccessible. The recent release of nanochat - a family of small LLMs with fully open pre-training data - addresses this as it provides a transparent view into where a model's parametric knowledge comes from. Towards the goal of understanding how knowledge is encoded by LLMs, we release NanoKnow, a
benchmark dataset that partitions questions from Natural Questions and SQuAD into splits based on whether their answers are present in nanochat's pre-training corpus. Using these splits, we can now properly disentangle the sources of knowledge that LLMs rely on when producing an output. To demonstrate NanoKnow's utility, we conduct experiments using eight nanochat checkpoints. Our findings show: (1) closed-book accuracy is strongly influenced by answer frequency in the pre-training data, (2) providing external evidence can mitigate this frequency dependence, (3) even with external evidence, models are more accurate when answers were seen during pre-training, demonstrating that parametric and external knowledge are complementary, and (4) non-relevant information is harmful, with accuracy decreasing based on both the position and the number of non-relevant contexts. We release all NanoKnow artifacts at https://github.com/castorini/NanoKnow.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.AI
ReSyn: Autonomously Scaling Synthetic Environments for Reasoning Models
Andre He, Nathaniel Weir, Kaj Bostrom et al. (+4 more)
Reinforcement learning with verifiable rewards (RLVR) has emerged as a promising approach for training
reasoning language models (RLMs) by leveraging supervision from verifiers. Although verifier implementation is easier than solution annotation for many tasks, existing synthetic data generation methods remain largely solution-centric, while verifier-based methods rely on a few hand-crafted procedural environments. In this work, we scale RLVR by introducing ReSyn, a pipeline that generates diverse
reasoning environments equipped with instance generators and verifiers, covering tasks such as constraint satisfaction, algorithmic puzzles, and spatial
reasoning. A Qwen2.5-7B-Instruct model trained with RL on ReSyn data achieves consistent gains across
reasoning benchmarks and out-of-domain math
benchmarks, including a 27% relative improvement on the challenging BBEH
benchmark. Ablations show that verifier-based supervision and increased task diversity both contribute significantly, providing empirical evidence that generating
reasoning environments at scale can enhance
reasoning abilities in RLMs
Read the full abstract ▾
All authors Andre He, Nathaniel Weir, Kaj Bostrom, Allen Nie, Darion Cassel, Sam Bayless, Huzefa Rangwala
Machine Learning
Reliable Abstention under Adversarial Injections: Tight Lower Bounds and New Upper Bounds
Ezra Edelman and Surbhi Goel
We study online learning in the
adversarial injection model introduced by [Goel et al. 2017], where a stream of labeled examples is predominantly drawn i.i.d.\ from an unknown distribution $\mathcal{D}$, but may be interspersed with adversarially chosen instances without the learner knowing which rounds are
adversarial. Crucially, labels are always consistent with a fixed target concept (the clean-label setting). The learner is additionally allowed to abstain from predicting, and the total error counts the mistakes whenever the learner decides to predict and incorrect abstentions when it abstains on i.i.d.\ rounds. Perhaps surprisingly, prior work shows that oracle access to the underlying distribution yields $O(d^2 \log T)$ combined error for VC dimension $d$, while distribution-agnostic
algorithms achieve only $\tilde{O}(\sqrt{T})$ for restricted classes, leaving open whether this gap is fundamental. We resolve this question by proving a matching $\Omega(\sqrt{T})$ lower bound for VC dimension $1$, establishing a sharp separation between the two information regimes. On the algorithmic side, we introduce a potential-based framework driven by robust witnesses, small subsets of labeled examples that certify predictions while remaining resilient to
adversarial contamination. We instantiate this framework using two combinatorial dimensions: (1) inference dimension, yielding combined error $\tilde{O}(T^{1-1/k})$ for classes of inference dimension $k$, and (2) certificate dimension, a new relaxation we introduce. As an application, we show that halfspaces in $\mathbb{R}^2$ have certificate dimension $3$, obtaining the first distribution-agnostic bound of $\tilde{O}(T^{2/3})$ for this class. This is notable since [Blum et al. 2021] showed halfspaces are not robustly learnable under clean-label attacks without abstention.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.AI
Align When They Want, Complement When They Need! Human-Centered Ensembles for Adaptive Human-AI Collaboration
Hasan Amin, Ming Yin and Rajiv Khanna
In human-AI decision making, designing AI that complements human expertise has been a natural strategy to enhance human-AI collaboration, yet it often comes at the cost of decreased AI performance in areas of human strengths. This can inadvertently erode human trust and cause them to ignore AI advice precisely when it is most needed. Conversely, an aligned AI fosters trust yet risks reinforcing suboptimal human behavior and lowering human-AI team performance. In this paper, we start by identifying this fundamental tension between performance-boosting (i.e., complementarity) and trust-building (i.e.,
alignment) as an inherent limitation of the traditional approach for training a single AI model to assist human decision making. To overcome this, we introduce a novel human-centered adaptive AI ensemble that strategically toggles between two specialist AI models - the aligned model and the complementary model - based on contextual cues, using an elegantly simple yet provably near-optimal Rational Routing Shortcut mechanism. Comprehensive theoretical analyses elucidate why the adaptive AI ensemble is effective and when it yields maximum benefits. Moreover, experiments on both simulated and real-world data show that when humans are assisted by the adaptive AI ensemble in decision making, they can achieve significantly higher performance than when they are assisted by single AI models that are trained to either optimize for their independent performance or even the human-AI team performance.
Read the full abstract ▾
Machine Learning
BarrierSteer: LLM Safety via Learning Barrier Steering
Thanh Q. Tran, Arun Verma, Kiwan Wong et al. (+3 more)
Despite the strong performance of large
language models (LLMs) across diverse tasks, their susceptibility to
adversarial attacks and unsafe content generation remains a significant obstacle to deployment, particularly in high-stakes settings. Addressing this challenge requires safety mechanisms that are both practically effective and theoretically grounded. In this paper, we introduce BarrierSteer, a novel inference-time framework that improves response safety by
embedding learned nonlinear safety constraints directly into the model's latent representation space. BarrierSteer treats hidden-state safety classifiers as Control Barrier Functions (CBFs), enabling constraint-guided steering of unsafe latent trajectories during generation. By composing multiple safety constraints through efficient constraint merging without modifying the underlying LLM parameters, BarrierSteer preserves model utility. We provide theoretical results showing that applying CBFs in the latent space yields a principled, modular, and computationally efficient approach for steering with respect to learned safety constraints, with guarantees conditional on the learned barriers capturing the intended safety property. Our extensive experimental results across multiple model families and
datasets demonstrate that BarrierSteer substantially reduces
adversarial attack success rates and unsafe generations, outperforming the existing method. The code is available in our \href{https://github.com/thanhquangtran/BarrierSteer}{GitHub repository}.
Read the full abstract ▾
All authors Thanh Q. Tran, Arun Verma, Kiwan Wong, Bryan Kian Hsiang Low, Daniela Rus, Wei Xiao
Machine Learning · Cross-listed from cs.MA
Descent-Guided Policy Gradient for Scalable Cooperative Multi-Agent Learning
Shan Yang and Yang Liu
Scaling cooperative multi-
agent reinforcement learning (MARL) is fundamentally limited by cross-
agent noise. When
agents share a common reward, each
agent's learning signal is computed from a shared return that depends on all
agents, so the stochasticity of the other
agents enters the signal as cross-
agent noise that grows with $N$. Many engineering systems, such as cloud computing and power systems, have differentiable analytical models that prescribe efficient system states, providing a new reference beyond noisy shared returns. In this work, we propose Descent-Guided Policy Gradient (DG-PG), a framework that augments policy-gradient updates with a low-noise descent signal toward this reference. We prove that DG-PG reduces policy-gradient
estimator variance from $\mathcal{O}(N)$ to $\mathcal{O}(1)$, preserves the stationary points of the cooperative objective, and achieves
agent-independent sample complexity $\mathcal{O}(1/\epsilon^3)$. We evaluate DG-PG on cloud resource scheduling and power dispatch tasks, which involve discrete and continuous decisions, respectively. In both tasks, DG-PG converges within 10 episodes on average at every scale up to 1500 heterogeneous
agents.
Read the full abstract ▾
Machine Learning
Generative Modeling via Kernelized Stochastic Interpolants
Florentin Coeurdoux, Etienne Lempereur, Nathanaël Cuvelle-Magar et al. (+2 more)
We develop a kernel method for generative modeling within the
stochastic interpolant framework, replacing neural network training with linear systems. The drift of the generative SDE is $\hat b_t(x) = \nabla\phi(x)^\top\eta_t$, where $\eta_t \in \mathbb{R}^P$ solves a $P\times P$ system computable from data, with $P$ independent of the data dimension $d$. Since estimates are inexact, the
diffusion coefficient $D_t$ affects sample quality; the optimal $D_t^*$ from Girsanov diverges at $t=0$, but this poses no difficulty and we develop an integrator that handles it seamlessly. The framework accommodates diverse feature maps: scattering transforms, pretrained generative models, etc, enabling generation and model combination without neural network training. We demonstrate the approach on financial time series,
turbulence, and image generation.
Read the full abstract ▾
All authors Florentin Coeurdoux, Etienne Lempereur, Nathanaël Cuvelle-Magar, Stéphane Mallat, Eric Vanden-Eijnden
Machine Learning · Cross-listed from cs.CV
The Invisible Gorilla Effect in Out-of-distribution Detection
Harry Anthony, Ziyun Liang, Hermione Warr et al. (+1 more)
Deep Neural Networks achieve high performance in
vision tasks by learning features from regions of interest (ROI) within images, but their performance degrades when deployed on out-of-distribution (OOD) data that differs from training data. This challenge has led to OOD detection methods that aim to identify and reject unreliable predictions. Although prior work shows that OOD detection performance varies by artefact type, the underlying causes remain underexplored. To this end, we identify a previously unreported bias in OOD detection: for hard-to-detect artefacts (near-OOD), detection performance typically improves when the artefact shares visual similarity (e.g. colour) with the model's ROI and drops when it does not - a phenomenon we term the Invisible Gorilla Effect. For example, in a skin lesion classifier with red lesion ROI, we show the method Mahalanobis Score achieves a 31.5% higher AUROC when detecting OOD red ink (similar to ROI) compared to black ink (dissimilar) annotations. We annotated artefacts by colour in 11,355 images from three public
datasets (e.g. ISIC) and generated colour-swapped counterfactuals to rule out
dataset bias. We then evaluated 40 OOD methods across 7
benchmarks and found significant performance drops for most methods when artefacts differed from the ROI. Our findings highlight an overlooked failure mode in OOD detection and provide guidance for more robust detectors. Code and annotations are available at: https://github.com/HarryAnthony/Invisible_Gorilla_Effect.
Read the full abstract ▾
All authors Harry Anthony, Ziyun Liang, Hermione Warr, Konstantinos Kamnitsas
Machine Learning
A Theory of How Pretraining Shapes Inductive Bias in Fine-Tuning
Nicolas Anguita, Francesco Locatello, Andrew M. Saxe et al. (+4 more)
Pretraining and fine-tuning are central stages in modern machine learning systems. In practice, feature learning plays an important role across both stages: deep neural networks learn a broad range of useful features during pretraining and further refine those features during fine-tuning. However, an end-to-end theoretical understanding of how choices of initialization impact the ability to reuse and refine features during fine-tuning has remained elusive. Here we develop an analytical theory of the pretraining fine-tuning pipeline in diagonal linear networks, deriving exact expressions for the generalization error as a function of initialization parameters and task statistics. We find that different initialization choices place the network into four distinct fine-tuning regimes that are distinguished by their ability to support feature learning and reuse and therefore by the task statistics for which they are beneficial. In particular, a smaller initialization scale in earlier layers enables the network to both reuse and refine its features, leading to superior generalization on fine-tuning tasks that rely on a subset of pretraining features. We demonstrate empirically that the same initialization parameters impact generalization in ResNets trained on CIFAR-100 and SVHN as well as
Transformers trained on modular arithmetic tasks. Overall, our results demonstrate an alytically how data and network initialization interact to shape fine-tuning generalization, highlighting an important role for the relative scale of initialization across different layers in enabling continued feature learning during fine-tuning.
Read the full abstract ▾
All authors Nicolas Anguita, Francesco Locatello, Andrew M. Saxe, Marco Mondelli, Flavia Mancini, Samuel Lippl, Clementine Domine
Machine Learning · Cross-listed from cs.CV
Closing the gap in multimodal medical representation alignment
Eleonora Grassucci, Giordano Cicchetti and Danilo Comminiello
In
multimodal learning, CLIP has emerged as the de-facto approach for mapping different modalities into a shared latent space by bringing semantically similar representations closer while pushing apart dissimilar ones. However, CLIP-based contrastive losses exhibit unintended behaviors that negatively impact true semantic
alignment, leading to sparse and fragmented latent spaces. This phenomenon, known as the modality gap, has been partially mitigated for standard text and image pairs but remains unknown and unresolved in more complex
multimodal settings, such as the medical domain. In this work, we study this phenomenon in the latter case, revealing that the modality gap is present also in medical
alignment, and we propose a modality-agnostic framework that closes this gap, ensuring that semantically related representations are more aligned, regardless of their source modality. Our method enhances
alignment between radiology images and clinical text, improving cross-modal
retrieval and image captioning.
Read the full abstract ▾
Machine Learning · Cross-listed from cs.AI
Latent Introspection: Models Can Detect Prior Concept Injections
Theia Pearson-Vogel, Martin Vanek, Raymond Douglas et al. (+1 more)
We uncover a latent capacity for introspection in a Qwen 32B model, demonstrating that the model can detect when concepts have been injected into its earlier context and identify which concept was injected. While the model denies injection in sampled outputs, logit lens analysis reveals clear detection signals in the residual stream, which are attenuated in the final layers. Furthermore, prompting the model with accurate information about AI introspection mechanisms can dramatically strengthen this effect: the sensitivity to injection increases massively (0.3% -> 39.9%) with only a 0.6% increase in false positives. Also, mutual information between nine injected and recovered concepts rises from 0.61 bits to 1.05 bits, ruling out generic noise explanations. Our results demonstrate models can have a surprising capacity for introspection and steering awareness that is easy to overlook, with consequences for latent
reasoning and safety.
Read the full abstract ▾
All authors Theia Pearson-Vogel, Martin Vanek, Raymond Douglas, Jan Kulveit
Machine Learning
Learning Discriminative and Generalizable Anomaly Detector for Dynamic Graph with Limited Supervision
Yuxing Tian, Yiyan Qi, Fengran Mo et al. (+3 more)
Dynamic graph anomaly detection is critical for many real-world applications but remains challenging due to the scarcity of labeled anomalies. Existing methods are either unsupervised or semi-supervised: unsupervised methods avoid the need for labeled anomalies but often produce ambiguous boundary, whereas semi-supervised methods can overfit to the limited labeled anomalies and generalize poorly to unseen anomalies. To address this gap, we consider a largely underexplored problem: learning a discriminative boundary from normal/unlabeled data, while leveraging limited labeled anomalies when available without sacrificing generalization to unseen anomalies. In this paper, we propose an effective, generalizable, and model-agnostic framework with three main components: (i) residual representation encoding that capture deviations between current interactions and their historical context, providing anomaly-relevant signals; (ii) a restriction loss that constrain the normal representations within an interval bounded by two co-centered hyperspheres, ensuring consistent scales while keeping anomalies separable; (iii) a bi-boundary optimization strategy that learns a discriminative and robust boundary using the log-likelihood distribution modeled by a normalizing flow. Extensive experiments demonstrate the superiority of our framework across diverse evaluation settings.
Read the full abstract ▾
All authors Yuxing Tian, Yiyan Qi, Fengran Mo, Weixu Zhang, Jian Guo, Jian-Yun Nie
Machine Learning
A Secure and Private Distributed Bayesian Federated Learning Design
Nuocheng Yang, Sihua Wang, Zhaohui Yang et al. (+3 more)
Distributed
Federated Learning (DFL) enables decentralized model training across large-scale systems without a central parameter server. However, DFL faces three critical challenges:
privacy leakage from honest-but-curious neighbors, slow convergence due to the lack of central coordination, and vulnerability to Byzantine adversaries aiming to degrade model accuracy. To address these issues, we propose a novel DFL framework that integrates Byzantine robustness,
privacy preservation, and convergence acceleration. Within this framework, each device trains a local model using a
Bayesian approach and independently selects an optimal subset of neighbors for posterior exchange. We formulate this neighbor selection as an optimization problem to minimize the global loss function under security and
privacy constraints. Solving this problem is challenging because devices only possess partial network information, and the complex coupling between
topology, security, and convergence remains unclear. To bridge this gap, we first analytically characterize the trade-offs between dynamic connectivity, Byzantine detection,
privacy levels, and convergence speed. Leveraging these insights, we develop a fully distributed Graph Neural Network (GNN)-based Reinforcement Learning (RL)
algorithm. This approach enables devices to make autonomous connection decisions based on local observations. Simulation results demonstrate that our method achieves superior robustness and efficiency with significantly lower overhead compared to traditional security and
privacy schemes.
Read the full abstract ▾
All authors Nuocheng Yang, Sihua Wang, Zhaohui Yang, Mingzhe Chen, Changchuan Yin, Kaibin Huang
Machine Learning · Cross-listed from cs.IR
FairFS: Addressing Deep Feature Selection Biases for Recommender System
Xianquan Wang, Zhaocheng Du, Jieming Zhu et al. (+3 more)
Large-scale online marketplaces and recommender systems serve as critical technological support for e-commerce development. In industrial recommender systems, features play vital roles as they carry information for downstream models. Accurate feature importance estimation is critical because it helps identify the most useful feature subsets from thousands of feature candidates for online services. Such selection enables improved online performance while reducing computational cost. To address feature selection problems in deep learning, trainable gate-based and sensitivity-based methods have been proposed and proven effective in industrial practice. However, through the analysis of real-world cases, we identified three bias issues that cause feature importance estimation to rely on partial model layers, samples, or gradients, ultimately leading to inaccurate importance estimation. We refer to these as layer bias, baseline bias, and approximation bias. To mitigate these issues, we propose FairFS, a fair and accurate feature selection
algorithm. FairFS regularizes feature importance estimated across all nonlinear transformation layers to address layer bias. It also introduces a smooth baseline feature close to the classifier decision boundary and adopts an aggregated approximation method to alleviate baseline and approximation biases. Extensive experiments demonstrate that FairFS effectively mitigates these biases and achieves state-of-the-art feature selection performance.
Read the full abstract ▾
All authors Xianquan Wang, Zhaocheng Du, Jieming Zhu, Qinglin Jia, Zhenhua Dong, Kai Zhang
Machine Learning
Counterfactual Understanding via Retrieval-aware Multimodal Modeling for Time-to-Event Survival Prediction
Ha-Anh Hoang Nguyen, Tri-Duc Phan Le, Duc-Hoang Pham et al. (+4 more)
This paper tackles the problem of time-to-event counterfactual survival prediction, aiming to optimize individualized survival outcomes in the presence of heterogeneity and censored data. We propose CURE, a framework that advances counterfactual survival modeling via comprehensive
multimodal embedding and latent subgroup
retrieval. CURE integrates clinical, paraclinical, demographic, and multi-omics information, which are aligned and fused through cross-
attention mechanisms. Complex multi-omics signals can be adaptively refined using a mixture-of-experts architecture, emphasizing the most informative omics components. Building upon this representation, CURE implicitly retrieves patient-specific latent subgroups that capture both baseline survival dynamics and treatment-dependent variations. Experimental results on METABRIC and TCGA-LUAD
datasets demonstrate that proposed CURE model consistently outperforms strong baselines in survival analysis, evaluated using the Time-dependent Concordance Index ($C^{td}$) and Integrated Brier Score (IBS). These findings highlight the potential of CURE to enhance
multimodal understanding and serve as a foundation for future treatment recommendation models. All code and related resources are publicly available to facilitate the reproducibility https://github.com/L2R-UET/CURE.
Read the full abstract ▾
All authors Ha-Anh Hoang Nguyen, Tri-Duc Phan Le, Duc-Hoang Pham, Huy-Son Nguyen, Cam-Van Thi Nguyen, Duc-Trong Le, Hoang-Quynh Le
Machine Learning · Cross-listed from astro-ph.IM
Multivariate time-series forecasting of ASTRI-Horn monitoring data: A Normal Behavior Model
Federico Incardona, Alessandro Costa, Farida Farsian et al. (+8 more)
This study presents a Normal Behavior Model (NBM) developed to forecast monitoring time-series data from the ASTRI-Horn Cherenkov
telescope under normal operating conditions. The analysis focused on 15 physical variables acquired by the
Telescope Control Unit between September 2022 and July 2024, representing sensor measurements from the Azimuth and Elevation motors. After data cleaning, resampling, feature selection, and correlation analysis, the
dataset was segmented into fixed-length intervals, in which the first I samples represented the input sequence provided to the model, while the forecast length, T, indicated the number of future time steps to be predicted. A sliding-window technique was then applied to increase the number of intervals. A Multi-Layer Perceptron (MLP) was trained to perform multivariate
forecasting across all features simultaneously. Model performance was evaluated using the Mean Squared Error (MSE) and the Normalized Median Absolute Deviation (NMAD), and it was also benchmarked against a Long Short-Term
Memory (LSTM) network. The MLP model demonstrated consistent results across different features and I-T configurations, and matched the performance of the LSTM while converging faster. It achieved an MSE of 0.019+/-0.003 and an NMAD of 0.032+/-0.009 on the test set under its best configuration (4 hidden layers, 720 units per layer, and I-T lengths of 300 samples each, corresponding to 5 hours at 1-minute resolution). Extending the forecast horizon up to 6.5 hours-the maximum allowed by this configuration-did not degrade performance, confirming the model's effectiveness in providing reliable hour-scale predictions. The proposed NBM provides a powerful tool for enabling early anomaly detection in online ASTRI-Horn monitoring time series, offering a basis for the future development of a prognostics and health management system that supports predictive maintenance.
Read the full abstract ▾
All authors Federico Incardona, Alessandro Costa, Farida Farsian, Francesco Franchina, Giuseppe Leto, Emilio Mastriani, Kevin Munari, Giovanni Pareschi, Salvatore Scuderi, Sebastiano Spinello, Gino Tosti
Machine Learning
A Computationally Efficient Multidimensional Vision Transformer
Alaa El Ichi and Khalide Jbilou
Vision Transformers have achieved state-of-the-art performance in a wide range of computer
vision tasks, but their practical deployment is limited by high computational and
memory costs. In this paper, we introduce a novel tensor-based framework for
Vision Transformers built upon the Tensor Cosine Product (Cproduct). By exploiting multilinear structures inherent in image data and the orthogonality of cosine transforms, the proposed approach enables efficient
attention mechanisms and structured feature representations. We develop the theoretical foundations of the tensor cosine product, analyze its algebraic properties, and integrate it into a new Cproduct-based
Vision Transformer architecture (TCP-ViT). Numerical experiments on standard classification and
segmentation benchmarks demonstrate that the proposed method achieves a uniform 1/C parameter reduction (where C is the number of channels) while maintaining competitive accuracy.
Read the full abstract ▾
Machine Learning
Discrete Diffusion Models Exploit Asymmetry to Solve Lookahead Planning Tasks
Itamar Trainin, Shauli Ravfogel, Omri Abend et al. (+1 more)
Discrete
diffusion models are often said to outperform autoregressive models on planning tasks, but the specific mechanisms enabling this are not yet well understood. To explain why, we abstract the complexity of studying "planning" into graph traversal problems, which offer a minimal setup for evaluating
language models' planning and lookahead capabilities. We find that non-autoregressive
language models are able to leverage an inherent directional asymmetry in lookahead planning tasks to outperform the training efficiency of their autoregressive counterparts. Through a mechanistic study of the training and inference dynamics of autoregressive and non-autoregressive models, we discover that while forward traversal through junctions in the graph requires complex sequential planning, the reverse path remains deterministic. Consequently, non-autoregressive models naturally develop a fundamentally different solution strategy. Effectively, they learn a reverse-decoding pattern which reduces an $\ell^{th}$-order lookahead problem to a sequence of simple $1^{st}$-order transitions. Latent-space analysis further confirms these divergent strategies by demonstrating the distinct internal representations formed during generation. Ultimately, these findings highlight the advantages of non-autoregressive modeling while clarifying the inherent planning limitations induced by autoregressive architectures.
Read the full abstract ▾
All authors Itamar Trainin, Shauli Ravfogel, Omri Abend, Amir Feder
Machine Learning
Unlearning Noise in PINNs: A Selective Pruning Framework for PDE Inverse Problems
Yongsheng Chen, Yong Chen, Wei Guo et al. (+1 more)
Physics-informed neural networks (PINNs) provide a promising framework for solving inverse problems governed by partial differential equations (PDEs) by integrating observational data and physical constraints in a unified optimization objective. However, the ill-posed nature of PDE inverse problems makes them highly sensitive to noise. Even a small fraction of corrupted observations can distort internal neural representations, severely impairing accuracy and destabilizing training. Motivated by recent advances in machine unlearning and structured network
pruning, we propose P-PINN, a selective
pruning framework designed to unlearn the influence of corrupted data in a pretrained PINN. Specifically, starting from a PINN trained on the full
dataset, P-PINN evaluates a joint residual–data fidelity indicator, a weighted combination of data misfit and PDE residuals, to partition the training set into reliable and corrupted subsets. Next, we introduce a bias-based
neuron importance measure that quantifies directional activation discrepancies between the two subsets, identifying
neurons whose representations are predominantly driven by corrupted samples. Building on this, an iterative
pruning strategy then removes noise-sensitive
neurons layer by layer. The resulting pruned network is fine-tuned on the reliable data subject to the original PDE constraints, acting as a lightweight post-processing stage rather than a complete retraining. Numerical experiments on extensive PDE inverse-problem
benchmarks demonstrate that P-PINN substantially improves robustness, accuracy, and training stability under noisy conditions, achieving up to a 96.6% reduction in relative error compared with baseline PINNs. These results indicate that activation-level post hoc
pruning is a promising mechanism for enhancing the reliability of physics-informed learning in noise-contaminated settings.
Read the full abstract ▾
All authors Yongsheng Chen, Yong Chen, Wei Guo, Xinghui Zhong
Machine Learning
On the Equivalence of Random Network Distillation, Deep Ensembles, and Bayesian Inference
Moritz A. Zanger, Yijun Wu, Pascal R. Van der Vaart et al. (+2 more)
Uncertainty quantification is central to safe and efficient deployments of deep learning models, yet many computationally practical methods lack lacking rigorous theoretical motivation. Random network
distillation (RND) is a lightweight technique that measures novelty via prediction errors against a fixed random target. While empirically effective, it has remained unclear what uncertainties RND measures and how its estimates relate to other approaches, e.g.
Bayesian inference or deep ensembles. This paper establishes these missing theoretical connections by analyzing RND within the neural tangent kernel framework in the limit of infinite network width. Our analysis reveals two central findings in this limit: (1) The uncertainty signal from RND – its squared self-predictive error – is equivalent to the predictive variance of a deep ensemble. (2) By constructing a specific RND target function, we show that the RND error distribution can be made to mirror the centered posterior predictive distribution of
Bayesian inference with wide neural networks. Based on this equivalence, we moreover devise a posterior sampling
algorithm that generates i.i.d. samples from an exact
Bayesian posterior predictive distribution using this modified
Bayesian RND model. Collectively, our findings provide a unified theoretical perspective that places RND within the principled frameworks of deep ensembles and
Bayesian inference, and offer new avenues for efficient yet theoretically grounded uncertainty quantification methods.
Read the full abstract ▾
All authors Moritz A. Zanger, Yijun Wu, Pascal R. Van der Vaart, Wendelin Böhmer, Matthijs T. J. Spaan
Machine Learning
Sparse Masked Attention Policies for Reliable Generalization
Caroline Horsch, Laurens Engwegen, Max Weltevrede et al. (+2 more)
In reinforcement learning, abstraction methods that remove unnecessary information from the observation are commonly used to learn policies which generalize better to unseen tasks. However, these methods often overlook a crucial weakness: the function which extracts the reduced-information representation has unknown generalization ability in unseen observations. In this paper, we address this problem by presenting an information removal method which more reliably generalizes to new states. We accomplish this by using a learned masking function which operates on, and is integrated with, the
attention weights within an
attention-based policy network. We demonstrate that our method significantly improves policy generalization to unseen tasks in the Procgen
benchmark compared to standard PPO and masking approaches.
Read the full abstract ▾
All authors Caroline Horsch, Laurens Engwegen, Max Weltevrede, Matthijs T. J. Spaan, Wendelin Böhmer
Machine Learning
DP-FedAdamW: An Efficient Optimizer for Differentially Private Federated Large Models
Jin Liu, Yinbin Miao, Ning Xi et al. (+1 more)
Balancing convergence efficiency and robustness under Differential
Privacy (DP) is a central challenge in
Federated Learning (FL). While AdamW accelerates training and fine-tuning in large-scale models, we find that directly applying it to Differentially Private FL (DPFL) suffers from three major issues: (i) data heterogeneity and
privacy noise jointly amplify the variance of second-moment
estimator, (ii) DP perturbations bias the second-moment
estimator, and (iii) DP amplify AdamW sensitivity to local overfitting, worsening client drift. We propose DP-FedAdamW, the first AdamW-based optimizer for DPFL. It restores AdamW under DP by stabilizing second-moment variance, removing DP-induced bias, and aligning local updates to the global descent to curb client drift. Theoretically, we establish an unbiased second-moment
estimator and prove a linearly accelerated convergence rate without any heterogeneity assumption, while providing tighter $(\varepsilon,\delta)$-DP guarantees. Our empirical results demonstrate the effectiveness of DP-FedAdamW across
language and
vision Transformers and ResNet-18. On Tiny-ImageNet (Swin-Base, $\varepsilon=1$), DP-FedAdamW outperforms the state-of-the-art (SOTA) by 5.83%. The code is available in Appendix.
Read the full abstract ▾
All authors Jin Liu, Yinbin Miao, Ning Xi, Junkang Liu
Machine Learning
A Replicate-and-Quantize Strategy for Plug-and-Play Load Balancing of Sparse Mixture-of-Experts LLMs
Zijie Liu, Jie Peng, Jinhao Duan et al. (+7 more)
Sparse Mixture-of-Experts (SMoE) architectures are increasingly used to scale large
language models efficiently, delivering strong accuracy under fixed compute budgets. However, SMoE models often suffer from severe load imbalance across experts, where a small subset of experts receives most
tokens while others are underutilized. Prior work has focused mainly on training-time solutions such as routing regularization or auxiliary losses, leaving inference-time behavior, which is critical for deployment, less explored. We present a systematic analysis of expert routing during inference and identify three findings: (i) load imbalance persists and worsens with larger batch sizes, (ii) selection frequency does not reliably reflect expert importance, and (iii) overall expert workload and importance can be estimated using a small calibration set. These insights motivate inference-time mechanisms that rebalance workloads without retraining or router modification. We propose Replicate-and-Quantize (R&Q), a training-free and near-lossless framework for dynamic workload rebalancing. In each layer, heavy-hitter experts are replicated to increase parallel capacity, while less critical experts and replicas are quantized to remain within the original
memory budget. We also introduce a Load-Imbalance Score (LIS) to measure routing skew by comparing heavy-hitter load to an equal allocation baseline. Experiments across representative SMoE models and
benchmarks show up to 1.4x reduction in imbalance with accuracy maintained within +/-0.6%, enabling more predictable and efficient inference.
Read the full abstract ▾
All authors Zijie Liu, Jie Peng, Jinhao Duan, Zirui Liu, Kaixiong Zhou, Mingfu Liang, Luke Simon, Xi Liu, Zhaozhuo Xu, Tianlong Chen