Feiyang Pan, Shenghe Zheng, Chunyan Yin, Guangbin Dou

MEMS gyroscopes play a critical role in inertial navigation and motion control applications but typically suffer from a fundamental trade-off between measurement range and noise performance. Existing hardware-based solutions aimed at mitigating this issue introduce additional complexity, cost, and scalability challenges. Deep-learning methods primarily focus on noise reduction and typically require precisely aligned ground-truth signals, making them difficult to deploy in practical scenarios and leaving the fundamental trade-off unresolved. To address these challenges, we introduce Mixture of Experts for MEMS Gyroscopes (MoE-Gyro), a novel self-supervised framework specifically designed for simultaneous over-range signal reconstruction and noise suppression. MoE-Gyro employs two experts: an Over‑Range Reconstruction Expert (ORE), featuring a Gaussian-Decay Attention mechanism for reconstructing saturated segments; and a Denoise Expert (DE), utilizing dual-branch complementary masking combined with FFT-guided augmentation for robust noise reduction. A lightweight gating module dynamically routes input segments to the appropriate expert. Furthermore, existing evaluation lack a comprehensive standard for assessing multi-dimensional signal enhancement. To bridge this gap, we introduce IMU Signal Enhancement Benchmark (ISEBench), an open-source benchmarking platform comprising the GyroPeak-100 dataset and a unified evaluation of IMU signal enhancement methods. We evaluate MoE-Gyro using our proposed ISEBench, demonstrating that our framework significantly extends the measurable range from ±450°/s to ±1500°/s, reduces Bias Instability by 98.4%, and achieves state-of-the-art performance, effectively addressing the long-standing trade-off in inertial sensing.Our code is available at: https://github.com/2002-Pan/Moe-Gyro

Rügamer, David, Liew, Bernard, Altai, Zainab, Stöcker, Almond

Semi-structured networks (SSNs) merge the structures familiar from additive models with deep neural networks, allowing the modeling of interpretable partial feature effects while capturing higher-order non-linearities at the same time. A significant challenge in this integration is maintaining the interpretability of the additive model component. Inspired by large-scale biomechanics datasets, this paper explores extending SSNs to functional data. Existing methods in functional data analysis are promising but often not expressive enough to account for all interactions and non-linearities and do not scale well to large datasets. Although the SSN approach presents a compelling potential solution, its adaptation to functional data remains complex. In this work, we propose a functional SSN method that retains the advantageous properties of classical functional regression approaches while also improving scalability. Our numerical experiments demonstrate that this approach accurately recovers underlying signals, enhances predictive performance, and performs favorably compared to competing methods.

Xi Chen, Chuan Qin, Ziqi Wang, Shasha Hu, Chao Wang, Hengshu Zhu, Hui Xiong

Open-set text classification (OSTC) requires models to correctly classify in-distribution (ID) samples while reliably rejecting out-of-distribution (OOD) inputs—an essential capability for real-world NLP systems. Most OSTC methods train on ID data under the closed assumption that all outputs belong to the known label space and then perform OOD detection with the biased representations, which inherently lack awareness of unknowns and thus yield overconfident predictions on OOD inputs. In this work, we present UnLLM, an Unknown-aware Large Language Model for OSTC. Instead of fixing classification to the entire known label space, we reformulate it into a subset-conditioned text generation task: the LLM is prompted with sampled subsets of known labels, and any instance outside the candidate set is explicitly assigned as “unknown”. This reformulation transforms OOD detection from a post-hoc procedure into an intrinsic modeling capability. More importantly, our approach is the first to explicitly incorporate the unknown into classification, enabling systematic modeling of unknowns through a unified representation–logits–inference optimization, which progressively strengthens the model’s capacity to capture open-set risk. Extensive experiments across six benchmarks show that UnLLM consistently outperforms state-of-the-art (SOTA) baselines. Code and datasets are available at https://github.com/cx9941/UnLLM.

Mentzelopoulos, Georgios, Chatzipantazis, Evangelos, Ramayya, Ashwin, Hedlund, Michelle, Buch, Vivek, Daniilidis, Kostas, Kording, Konrad, Vitale, Flavia

Deep learning based neural decoding from stereotactic electroencephalography (sEEG) would likely benefit from scaling up both dataset and model size. To achieve this, combining data across multiple subjects is crucial. However, in sEEG cohorts, each subject has a variable number of electrodes placed at distinct locations in their brain, solely based on clinical needs. Such heterogeneity in electrode number/placement poses a significant challenge for data integration, since there is no clear correspondence of the neural activity recorded at distinct sites between individuals. Here we introduce seegnificant: a training framework and architecture that can be used to decode behavior across subjects using sEEG data. We tokenize the neural activity within electrodes using convolutions and extract long-term temporal dependencies between tokens using self-attention in the time dimension. The 3D location of each electrode is then mixed with the tokens, followed by another self-attention in the electrode dimension to extract effective spatiotemporal neural representations. Subject-specific heads are then used for downstream decoding tasks. Using this approach, we construct a multi-subject model trained on the combined data from 21 subjects performing a behavioral task. We demonstrate that our model is able to decode the trial-wise response time of the subjects during the behavioral task solely from neural data. We also show that the neural representations learned by pretraining our model across individuals can be transferred in a few-shot manner to new subjects. This work introduces a scalable approach towards sEEG data integration for multi-subject model training, paving the way for cross-subject generalization for sEEG decoding.

Fu, Kexue, luo, xiaoyuan, Qu, Linhao, Wang, Shuo, Xiong, Ying, Maglogiannis, Ilias, Gao, Longxiang, Wang, Manning

The expensive fine-grained annotation and data scarcity have become the primary obstacles for the widespread adoption of deep learning-based Whole Slide Images (WSI) classification algorithms in clinical practice. Unlike few-shot learning methods in natural images that can leverage the labels of each image, existing few-shot WSI classification methods only utilize a small number of fine-grained labels or weakly supervised slide labels for training in order to avoid expensive fine-grained annotation. They lack sufficient mining of available WSIs, severely limiting WSI classification performance. To address the above issues, we propose a novel and efficient dual-tier few-shot learning paradigm for WSI classification, named FAST. FAST consists of a dual-level annotation strategy and a dual-branch classification framework. Firstly, to avoid expensive fine-grained annotation, we collect a very small number of WSIs at the slide level, and annotate an extremely small number of patches. Then, to fully mining the available WSIs, we use all the patches and available patch labels to build a cache branch, which utilizes the labeled patches to learn the labels of unlabeled patches and through knowledge retrieval for patch classification. In addition to the cache branch, we also construct a prior branch that includes learnable prompt vectors, using the text encoder of visual-language models for patch classification. Finally, we integrate the results from both branches to achieve WSI classification. Extensive experiments on binary and multi-class datasets demonstrate that our proposed method significantly surpasses existing few-shot classification methods and approaches the accuracy of fully supervised methods with only 0.22% annotation costs. All codes and models will be publicly available on https://github.com/fukexue/FAST.

Dikshant Shehmar, Matthew Schlegel, Matthew Taylor, Marlos C. Machado

Planning with a learned model remains a key challenge in model-based reinforcement learning (RL). In decision-time planning, state representations are critical as they must support local cost computation while preserving long-horizon structure. In this paper, we show that the Laplacian representation provides an effective latent space for planning by capturing state-space distances at multiple time scales. This representation preserves meaningful distances and naturally decomposes long-horizon problems into subgoals, also mitigating the compounding errors that arise over long prediction horizons. Building on these properties, we introduce ALPS, a hierarchical planning algorithm, and demonstrate that it outperforms commonly used model-free baselines on a selection of offline goal-conditioned RL tasks from OGBench.

WEI-KAI CHANG, Rajiv Khanna

As deep learning models continue to scale, the growing computational demands have amplified the need for effective coreset selection techniques. Coreset selection aims to accelerate training by identifying small, representative subsets of data that approximate the performance of the full dataset. Among various approaches, gradient-based methods stand out due to their strong theoretical underpinnings and practical benefits, particularly under limited data budgets. However, these methods face challenges such as naïve stochastic gradient descent (SGD) acting as a surprisingly strong baseline and the breakdown of representativeness due to loss curvature mismatches over time. In this work, we propose a novel framework that addresses these limitations. First, we establish a connection between posterior sampling and loss landscapes, enabling robust coreset selection even in high-data-corruption scenarios. Second, we introduce a smoothed loss function based on posterior sampling onto the model weights, enhancing stability and generalization while maintaining computational efficiency. We also present a novel convergence analysis for our sampling-based coreset selection method. Finally, through extensive experiments, we demonstrate how our approach achieves faster training and enhanced generalization across diverse datasets than the current state of the art.

Pascal Janetzky, Tobias Schlagenhauf, Stefan Feuerriegel

Continual learning (CL) aims to train models sequentially over multiple domains without forgetting previously learned knowledge. However, existing CL methods optimize for in-domain performance and are therefore prone to learning spurious, domain-specific cues ("shortcut learning"), which limits generalization to unseen domains after deployment. In this paper, we address this limitation through *continual learning of domain-invariant representation*. We introduce a broad class of CL methods that sequentially learn representations capturing invariant structures across domains. Our methods are motivated by the observation that such invariant structures often preserve the underlying causal mechanisms, which can reduce the risk of overfitting to domain-specific cues and thus offer better out-of-domain generalization. Our proposed CL methods combine replay-based training with a tailored sequential invariance alignment to learn---and preserve---invariant structures over time. We evaluate our methods under a deployment-oriented protocol that measures performance on unseen target domains. Across six benchmark and real-world datasets spanning vision, medicine, manufacturing, and ecology, our methods consistently outperform existing CL baselines in terms of generalization to unseen target domains. As an ablation, we further show that naïve extensions of sequential training with existing domain-invariant representation learning (DIRL) methods provide only limited benefits. To the best of our knowledge, this is the first work to develop domain-invariant representation methods for CL.

Jiraphon Yenphraphai, Ashkan Mirzaei, Jianqi Chen, Jiaxu Zou, Sergey Tulyakov, Raymond A. Yeh, Peter Wonka, Chaoyang Wang

Video-conditioned 4D shape generation aims to recover time-varying 3D geometry and view-consistent appearance directly from an input video. In this work, we introduce a native video-to-4D shape generation framework that synthesizes a single dynamic 3D representation end-to-end from the video. Our framework introduces three key components based on large-scale pre-trained 3D models: (i) a temporal attention that conditions generation on all frames while producing a time-indexed dynamic representation; (ii) a time-aware point sampling and 4D latent anchoring that promote temporally consistent geometry and texture; and (iii) noise sharing across frames to enhance temporal stability. Our method accurately captures non-rigid motion, volume changes, and even topological transitions without per-frame optimization. Across diverse in-the-wild videos, our method improves robustness and perceptual fidelity and reduces failure modes compared with the baselines.

Ye, Zhifan, Wan, Chenxi, Li, Chaojian, Hong, Jihoon, Li, Sixu, Li, Leshu, Zhang, Yongan, Lin, Yingyan (Celine)

3D Gaussian splatting has recently emerged as a promising technique for novel view synthesis from sparse image sets, yet comes at the cost of requiring millions of 3D Gaussian primitives to reconstruct each 3D scene. This largely limits its application to resource-constrained devices and applications.Despite advances in Gaussian pruning techniques that aim to remove individual 3D Gaussian primitives, the significant reduction in primitives often fails to translate into commensurate increases in rendering speed, impeding efficiency and practical deployment. We identify that this discrepancy arises due to the overlooked impact of fragment count per Gaussian (i.e., the number of pixels each Gaussian is projected onto). To bridge this gap and meet the growing demands for efficient on-device 3D Gaussian rendering, we propose fragment pruning, an orthogonal enhancement to existing pruning methods that can significantly accelerate rendering by selectively pruning fragments within each Gaussian. Our pruning framework dynamically optimizes the pruning threshold for each Gaussian, markedly improving rendering speed and quality. Extensive experiments in both static and dynamic scenes validate the effectiveness of our approach. For instance, by integrating our fragment pruning technique with state-of-the-art Gaussian pruning methods, we achieve up to a 1.71$\times$ speedup on an edge GPU device, the Jetson Orin NX, and enhance rendering quality by an average of 0.16 PSNR on the Tanks\&Temples dataset. Our code is available at https://github.com/GATECH-EIC/Fragment-Pruning.

Jazbec, Metod, Timans, Alexander, Hadži Veljković, Tin, Sakmann, Kaspar, Zhang, Dan, Andersson Naesseth, Christian, Nalisnick, Eric

Scaling machine learning models significantly improves their performance. However, such gains come at the cost of inference being slow and resource-intensive. Early-exit neural networks (EENNs) offer a promising solution: they accelerate inference by allowing intermediate layers to exit and produce a prediction early. Yet a fundamental issue with EENNs is how to determine when to exit without severely degrading performance. In other words, when is it 'safe' for an EENN to go 'fast'? To address this issue, we investigate how to adapt frameworks of risk control to EENNs. Risk control offers a distribution-free, post-hoc solution that tunes the EENN's exiting mechanism so that exits only occur when the output is of sufficient quality. We empirically validate our insights on a range of vision and language tasks, demonstrating that risk control can produce substantial computational savings, all the while preserving user-specified performance goals.

Gunho Park, Jeongin Bae, Byeongwook Kim, Baeseong Park, Jiwon Ryu, Hoseung Kim, Se Jung Kwon, Dongsoo Lee

Weight-only quantization is widely used to mitigate the memory-bound nature of LLM inference. Codebook-based methods extend this trend by achieving strong accuracy in the extremely low-bit regime (e.g., 2-bit). However, current kernels rely on dequantization, which repeatedly fetches centroids and reconstructs weights, incurring substantial latency and cache pressure. We present CodeGEMM, a codebook-centric GEMM kernel that replaces dequantization with precomputed inner products between centroids and activations stored in a lightweight Psumbook. At inference, code indices directly gather these partial sums, eliminating per-element lookups and reducing the on-chip footprint. The kernel supports the systematic exploration of latency–memory–accuracy trade-offs under a unified implementation. On Llama-3 models, CodeGEMM delivers 1.83x (8B) and 8.93x (70B) speedups in the 2-bit configuration compared to state-of-the-art codebook-based quantization at comparable accuracy and further improves computing efficiency and memory subsystem utilization.

Yixu Feng, Zinan Zhao, Yanxiang Ma, Chenghao Xia, Chengbin Du, Yunke Wang, Chang Xu

Vision-Language-Action (VLA) models have shown remarkable promise in robotics manipulation, yet their high computational cost hinders real-time deployment. Existing token pruning methods suffer from a fundamental trade-off: aggressive compression using pruning inevitably discards critical geometric details like contact points, leading to severe performance degradation. This forces a compromise, limiting the achievable compression rate and thus the potential speedup. We argue that breaking this trade-off requires rethinking compression as a geometry-aware, continuous token resampling in the vision encoder. To this end, we propose the *Differentiable Grid Sampler (GridS)*, a plug-and-play module that performs task-aware, continuous resampling of visual tokens in VLA. By adaptively predicting a minimal set of salient coordinates and extracting features via differentiable interpolation, GridS preserves essential spatial information while achieving drastic compression (with fewer than 10\% original visual tokens). Experiments on both LIBERO benchmark and a real robotic platform demonstrate that validating the lowest feasible visual token count reported to date, GridS achieves a 76\% reduction in FLOPs with no degradation in the success rate. The code is available at https://github.com/Fediory/Grid-Sampler.

Dong, Jiahua, Liang, Wenqi, Li, Hongliu, Zhang, Duzhen, Cao, Meng, Ding, Henghui, Khan, Salman H., Shahbaz Khan, Fahad

Custom diffusion models (CDMs) have attracted widespread attention due to their astonishing generative ability for personalized concepts. However, most existing CDMs unreasonably assume that personalized concepts are fixed and cannot change over time. Moreover, they heavily suffer from catastrophic forgetting and concept neglect on old personalized concepts when continually learning a series of new concepts. To address these challenges, we propose a novel Concept-Incremental text-to-image Diffusion Model (CIDM), which can resolve catastrophic forgetting and concept neglect to learn new customization tasks in a concept-incremental manner. Specifically, to surmount the catastrophic forgetting of old concepts, we develop a concept consolidation loss and an elastic weight aggregation module. They can explore task-specific and task-shared knowledge during training, and aggregate all low-rank weights of old concepts based on their contributions during inference. Moreover, in order to address concept neglect, we devise a context-controllable synthesis strategy that leverages expressive region features and noise estimation to control the contexts of generated images according to user conditions. Experiments validate that our CIDM surpasses existing custom diffusion models. The source codes are available at https://github.com/JiahuaDong/CIFC.

Shurong Wang, Yuqi Pan, Zhuoyang Shen, Meng Zhang, Hongwei Wang, Guoqi Li

Associative memory models are content-addressable memory systems fundamental to biological intelligence and are notable for their high interpretability. However, existing models evaluate the quality of retrieval based on proximity, which cannot guarantee that the retrieved pattern has the strongest association with the query, failing correctness. We reframe this problem by proposing that a query is a generative variant of a stored memory pattern, and define a variant distribution to model this subtle context-dependent generative process. Consequently, correct retrieval should return the memory pattern with the maximum a posteriori probability of being the query's origin. This perspective reveals that an ideal similarity measure should approximate the likelihood of each stored pattern generating the query in accordance with variant distribution, which is impossible for fixed and pre-defined similarities used by existing associative memories. To this end, we develop adaptive similarity, a novel mechanism that learns to approximate this insightful but unknown likelihood from samples drawn from context, aiming for correct retrieval. We theoretically prove that our proposed adaptive similarity achieves optimal correct retrieval under three canonical and widely applicable types of variants: noisy, masked, and biased. We integrate this mechanism into a novel adaptive Hopfield network (`A-Hop`), and empirical results show that it achieves state-of-the-art performance across diverse tasks, including memory retrieval, tabular classification, image classification, and multiple instance learning. Our code is publicly available at https://github.com/shurongwang/Adaptive-Hopfield-Network.

Gao, Weizhi, Hou, Zhichao, Xu, Han, Liu, Xiaorui

Implicit models such as Deep Equilibrium Models (DEQs) have emerged as promising alternative approaches for building deep neural networks. Their certified robustness has gained increasing research attention due to security concerns. Existing certified defenses for DEQs employing interval bound propagation and Lipschitz-bounds not only offer conservative certification bounds but also are restricted to specific forms of DEQs. In this paper, we provide the first randomized smoothing certified defense for DEQs to solve these limitations. Our study reveals that simply applying randomized smoothing to certify DEQs provides certified robustness generalized to large-scale datasets but incurs extremely expensive computation costs. To reduce computational redundancy, we propose a novel Serialized Randomized Smoothing (SRS) approach that leverages historical information. Additionally, we derive a new certified radius estimation for SRS to theoretically ensure the correctness of our algorithm. Extensive experiments and ablation studies on image recognition demonstrate that our algorithm can significantly accelerate the certification of DEQs by up to 7x almost without sacrificing the certified accuracy. The implementation will be publicly available upon the acceptance of this work. Our code is available at https://github.com/WeizhiGao/Serialized-Randomized-Smoothing.

Tim van Erven, Jack Mayo, Julia Olkhovskaya, Chen-Yu Wei

We present an efficient algorithm for linear contextual bandits with adversarial losses and stochastic action sets. Our approach reduces this setting to misspecification-robust adversarial linear bandits with fixed action sets. Without knowledge of the context distribution or access to a context simulator, the algorithm achieves $\tilde{O}(\min \{d^2\sqrt{T},\sqrt{d^{3}T\log K}\})$ regret and runs in $\mathrm{poly}(d,C,T)$ time, where $d$ is the feature dimension, $C$ is an upper bound on the number of linear constraints defining the action set in each round, $K$ is an upper bound on the number of actions in each round, and $T$ is number of rounds. This resolves the open question by Liu et al. (2023) on whether one can obtain $\mathrm{poly}(d)\sqrt{T}$ regret in polynomial time independent of the number of actions. For the important class of combinatorial bandits with adversarial losses and stochastic action sets where the action sets can be described by a polynomial number of linear constraints, our algorithm is the first to achieve $\mathrm{poly}(d)\sqrt{T}$ regret in polynomial time, while no prior algorithm achieves even $o(T)$ regret in polynomial time to our knowledge. When a simulator is available, the regret bound can be improved to $\tilde{O}(d\sqrt{L^{\star }})$, where $L^{\star }$ is the cumulative loss of the best policy.

Xijun Liu, Zhaoliang Zhang, Yuxiang Guo, Yifan Zhou, Rama Chellappa, Cheng Peng

Significant progress has been made in photo-realistic scene reconstruction over recent years. Various disparate efforts have enabled capabilities such as multi-appearance or large-scale reconstruction from images acquired by consumer-grade cameras. How far away are we from digitally replicating the real world in 4D? So far, there appears to be a lack of well-designed dataset that can evaluate the holistic progress on large-scale scene reconstruction. We introduce a collection of imagery on a campus, acquired at different seasons, times of day, from multiple elevations, views, and at scale. To estimate many camera poses over such a large area and across elevations, we apply a semi-automated calibration pipeline to eliminate visual ambiguities and avoid excessive matching, then visually verify all calibration results to ensure accuracy. Finally, we benchmark various algorithms for automatic calibration and dense reconstruction on our dataset, named ULTRA-360, and demonstrate numerous potential areas to improve upon, e.g., balancing sensitivity and specificity in feature matching, densification and floaters in dense reconstruction, multi-appearance overfitting, etc. We believe ULTRA-360 can serve as the benchmark that reflect realistic challenges in an end-to-end scene-reconstruction pipeline.

Kuratov, Yury, Bulatov, Aydar, Anokhin, Petr, Rodkin, Ivan, Sorokin, Dmitry, Sorokin, Artyom, Burtsev, Mikhail

In recent years, the input context sizes of large language models (LLMs) have increased dramatically. However, existing evaluation methods have not kept pace, failing to comprehensively assess the efficiency of models in handling long contexts. To bridge this gap, we introduce the BABILong benchmark, designed to test language models' ability to reason across facts distributed in extremely long documents. BABILong includes a diverse set of 20 reasoning tasks, including fact chaining, simple induction, deduction, counting, and handling lists/sets. These tasks are challenging on their own, and even more demanding when the required facts are scattered across long natural text. Our evaluations show that popular LLMs effectively utilize only 10-20% of the context and their performance declines sharply with increased reasoning complexity. Among alternatives to in-context reasoning, Retrieval-Augmented Generation methods achieve a modest 60% accuracy on single-fact question answering, independent of context length. Among context extension methods, the highest performance is demonstrated by recurrent memory transformers after fine-tuning, enabling the processing of lengths up to 50 million tokens. The BABILong benchmark is extendable to any length to support the evaluation of new upcoming models with increased capabilities, and we provide splits up to 10 million token lengths.

Julius Vetter, Manuel Gloeckler, Daniel Gedon, Jakob H Macke

Simulation-based inference (SBI) offers a flexible and general approach to performing Bayesian inference: In SBI, a neural network is trained on synthetic data simulated from a model and used to rapidly infer posterior distributions for observed data. A key goal for SBI is to achieve accurate inference with as few simulations as possible, especially for expensive simulators. In this work, we address this challenge by repurposing recent probabilistic foundation models for tabular data: We show how tabular foundation models---specifically TabPFN---can be used as pre-trained autoregressive conditional density estimators for SBI. We propose Neural Posterior Estimation with Prior-data Fitted Networks (NPE-PFN) and show that it is competitive with current SBI approaches in terms of accuracy for both benchmark tasks and two complex scientific inverse problems. Crucially, it often substantially outperforms them in terms of simulation efficiency, sometimes requiring orders of magnitude fewer simulations. NPE-PFN eliminates the need for selecting and training an inference network and tuning its hyperparameters. We also show that it exhibits superior robustness to model misspecification and can be scaled to simulation budgets that exceed the context size limit of TabPFN. NPE-PFN provides a new direction for SBI, where training-free, general-purpose inference models offer efficient, easy-to-use, and flexible solutions for a wide range of stochastic inverse problems.

Kunal Pratap Singh, Ali Garjani, Rishubh Singh, Muhammad Uzair Khattak, Jason Toskov, Efe Tarhan, Andrei Atanov, Oğuzhan Kar, Amir Zamir

The common approach for developing a vision model is generalism, which involves training on a large diverse dataset to cover the varied deployment environments and leads to a model that is expected to solve the problem everywhere. However, many practical applications need to operate in a specific test space, e.g., a robot deployed in a single house, and do not necessarily need to generalize to novel environments. In this work, we explore whether we can use rich multimodal data only from the test environment to pre-train a representation in a self-supervised way, without access to any external data. We find that this approach can match and, in most cases, outperform generalists pre-trained on large-scale Internet datasets, including popular off-the-shelf models, CLIP and DINOv2. We study the effectiveness of this approach by evaluating the models on various datasets and downstream tasks, such as semantic segmentation, captioning, and object detection, as well as a set of ablations and analyses to extract insights. This approach raises intriguing points on substituting data with (multi)modality, enabling an alternative scenario where the need for external Internet-scale datasets for pre-training models is reduced. It also shows that merely benefiting from test-space data was insufficient for achieving competitive results, and multimodality was essential for that purpose.

Sihan Yang, Runsen Xu, Yiman Xie, Sizhe Yang, Mo Li, Jingli Lin, Chenming Zhu, Xiaochen Chen, Haodong Duan, Xiangyu Yue, Dahua Lin, Tai Wang, Jiangmiao Pang

Spatial intelligence is essential for multimodal large language models (MLLMs) operating in the complex physical world. Existing benchmarks, however, probe only single-image relations and thus fail to assess the multi-image spatial reasoning that real-world deployments demand. We introduce MMSI-Bench, a VQA benchmark dedicated to multi-image spatial intelligence. Six 3D-vision researchers spent more than 300 hours meticulously crafting 1,000 challenging, unambiguous multiple-choice questions from over 120,000 images, each paired with carefully designed distractors and a stepwise reasoning process. We conduct extensive experiments and evaluate 37 open-source and proprietary MLLMs, observing a wide gap: the strongest open-source model attains roughly 30\% accuracy and OpenAI's GPT-5 reasoning model reaches 40\%, while humans score 97\%. These results underscore the challenging nature of MMSI-Bench and the substantial headroom for future research. Leveraging the annotated reasoning processes, we also provide an automated error analysis pipeline that diagnoses four dominant failure modes, including (1) grounding errors, (2) overlap-matching and scene-reconstruction errors, (3) situation-transformation reasoning errors, and (4) spatial-logic errors, offering insights for advancing spatial intelligence.

Fu, Shenghao, Yan, Junkai, Yang, Qize, Wei, Xihan, Xie, Xiaohua, Zheng, Wei-Shi

Recent vision foundation models can extract universal representations and show impressive abilities in various tasks. However, their application on object detection is largely overlooked, especially without fine-tuning them. In this work, we show that frozen foundation models can be a versatile feature enhancer, even though they are not pre-trained for object detection. Specifically, we explore directly transferring the high-level image understanding of foundation models to detectors in the following two ways. First, the class token in foundation models provides an in-depth understanding of the complex scene, which facilitates decoding object queries in the detector's decoder by providing a compact context. Additionally, the patch tokens in foundation models can enrich the features in the detector's encoder by providing semantic details. Utilizing frozen foundation models as plug-and-play modules rather than the commonly used backbone can significantly enhance the detector's performance while preventing the problems caused by the architecture discrepancy between the detector's backbone and the foundation model. With such a novel paradigm, we boost the SOTA query-based detector DINO from 49.0% AP to 51.9% AP (+2.9% AP) and further to 53.8% AP (+4.8% AP) by integrating one or two foundation models respectively, on the COCO validation set after training for 12 epochs with R50 as the detector's backbone. Code will be available.

Lucas Zoroddu, Pierre Humbert, Laurent Oudre

Expert-annotated time series datasets often suffer from low agreement, especially in medical applications where decisions rely on subjective criteria and inconsistent thresholds. Such variability degrades annotation quality and thus limits the reliability of supervised classification models. To address this, we propose to rely on a pairwise comparison-based approach, which provides a more robust alternative to individual annotation, since relative judgments are typically easier and yield higher consistency. The problem is thus transformed into a ranking problem and we introduce an ideal point model adapted to time series data using elastic similarity measures such as Dynamic Time Warping (DTW) and Time Warp Edit Distance (TWED). We prove Lipschitz continuity of these distances and demonstrate several convergence guarantees for this model. To facilitate gradient-based optimization, we also introduce a differentiable version of the TWED. Finally, we show through multiple experiments that our approach produces accurate and robust rankings under noisy annotation conditions.

Meshi, Ofer, Schwing, Alexander

Finding the maximum a-posteriori (MAP) assignment is a central task in graphical models. Since modern applications give rise to very large problem instances, there is increasing need for efficient solvers. In this work we propose to improve the efficiency of coordinate-minimization-based dual-decomposition solvers by running their updates asynchronously in parallel. In this case message-passing inference is performed by multiple processing units simultaneously without coordination, all reading and writing to shared memory. We analyze the convergence properties of the resulting algorithms and identify settings where speedup gains can be expected. Our numerical evaluations show that this approach indeed achieves significant speedups in common computer vision tasks.

Eleonora Grassucci, Giordano Cicchetti, Emanuele Frasca, Aurelio Uncini, Danilo Comminiello

In multimodal learning, CLIP has been recognized as the \textit{de facto} method for learning a shared latent space across multiple modalities, placing similar representations close to each other and moving them away from dissimilar ones. Although CLIP-based losses effectively align modalities at the semantic level, the resulting latent spaces often remain only partially shared, revealing a structural mismatch known as the modality gap. While the necessity of addressing this phenomenon remains debated, particularly given its limited impact on instance-wise tasks (e.g., retrieval), we prove that its influence is more pronounced in group-level tasks (e.g., clustering). To support this claim, we introduce a novel method designed to consistently reduce this discrepancy in two-modal settings, with a straightforward extension to the general $n$-modal case. Through our extensive evaluation, we prove our novel insight: while reducing the gap provides only marginal or inconsistent improvements in traditional instance-wise tasks, it significantly enhances group-wise tasks. These findings may reshape our understanding of the modality gap, highlighting its key role in improving performance on tasks requiring semantic grouping.

Joy Dhar, Song Xia, Manish Pandey, Maryam Haghighat, Azadeh Alavi, Ferdous Sohel, Wenyu Zhang, Nayyar Zaidi

We introduce Hybrid Convolutions with Attention Stochasticity (HyCAS), an adversarial defense that narrows the long-standing gap between provable robustness under ℓ2 certificates and empirical robustness against strong ℓ∞ attacks, while preserving strong generalization across diverse imaging benchmarks. HyCAS unifies deterministic and randomized principles by coupling 1-Lipschitz, spectrally normalized convolutions with two stochastic components—spectral normalized random-projection filters and a randomized attention-noise mechanism—to realize a randomized defense. Injecting smoothing randomness inside the architecture yields an overall ≤ 2-Lipschitz network with formal certificates. Extensive experiments on diverse imaging benchmarks—including CIFAR-10/100, ImageNet-1k, NIH Chest X-ray, HAM10000—show that HyCAS surpasses prior leading certified and empirical defenses, boosting certified accuracy by up to ≈ 7.3% (on NIH Chest X-ray) and empirical robustness by up to ≈ 3.1% (on HAM10000), without sacrificing clean accuracy. These results show that a randomized Lipschitz constrained architecture can simultaneously improve both certified ℓ2 and empirical ℓ∞ adversarial robustness, thereby supporting safer deployment of deep models in high-stakes applications.

Zheng Qi, Chao Shang, Evangelia Spiliopoulou, Nikolaos Pappas

Vision language models (VLMs) often generate hallucination, i.e., content that cannot be substantiated by either textual or visual inputs. Prior work primarily attributes this to over-reliance on linguistic prior knowledge rather than visual inputs. Some methods attempt to mitigate hallucination by amplifying visual token attention proportionally to their attention scores. However, these methods overlook the visual attention sink problem, where attention is frequently misallocated to task-irrelevant visual regions, and neglect cross-modal fusion balance by enhancing only visual attention without adjusting attention to the user query. This can result in amplifying incorrect areas while failing to properly interpret the user query. To address these challenges, we propose a simple yet effective method called Gaze Shift-Guided Cross-modal Fusion Enhancement (GIFT). GIFT pre-computes a holistic visual saliency map by tracking positive changes in visual attention, or "gaze shifts", during user query comprehension, and leverages this map to amplify attention to both salient visual information and the user query at each decoding step. This reduces the impact of visual attention sink, as irrelevant tokens exhibit minimal shifts, while ensuring balanced cross-modal fusion for well-integrated representation. Extensive experiments show that GIFT effectively mitigates hallucination in VLMs across both generative and classification tasks, achieving up to 20.7% improvement over greedy decoding, while maintaining general vision-language performance with low computational overhead.

Mészáros, Anna, Ujváry, Szilvia, Brendel, Wieland, Reizinger, Patrik, Huszar, Ferenc

LLMs show remarkable emergent abilities, such as inferring concepts from presumably out-of-distribution prompts, known as in-context learning. Though this success is often attributed to the Transformer architecture, our systematic understanding is limited. In complex real-world data sets, even defining what is out-of-distribution is not obvious. To better understand the OOD behaviour of autoregressive LLMs, we focus on formal languages, which are defined by the intersection of rules. We define a new scenario of OOD compositional generalization, termed \textit{rule extrapolation}. Rule extrapolation describes OOD scenarios, where the prompt violates at least one rule. We evaluate rule extrapolation in formal languages with varying complexity in linear and recurrent architectures, the Transformer, and state space models to understand the architectures' influence on rule extrapolation. We also lay the first stones of a normative theory of rule extrapolation, inspired by the Solomonoff prior in algorithmic information theory.

Weiyun Wang, Zhangwei Gao, Lianjie Chen, Zhe Chen, Jinguo Zhu, Xiangyu Zhao, Yangzhou Liu, Yue Cao, Shenglong Ye, Xizhou Zhu, Lewei Lu, Haodong Duan, Yu Qiao, Jifeng Dai, Wenhai Wang

We construct VisualPRM400K, a dataset comprising about 400K multimodal process supervision data. Building upon this dataset, we develop VisualPRM, an advanced multimodal Process Reward Model (PRM) capable of estimating the value score of each step during the reasoning process. Under the Best-of-N evaluation setting, our model improves the reasoning performance of three types of MLLMs and four different model scales. Even when applied to the highly capable InternVL2.5-78B, it achieves a 5.9-point improvement across seven multimodal reasoning benchmarks. Experimental results show that the PRM model trained on our VisualPRM400K exhibits superior performance compared to Outcome Reward Models and Self-Consistency during BoN evaluation. To further facilitate the development of multimodal PRMs, we construct VisualProcessBench, a benchmark designed to measure the abilities of PRMs and MLLMs to detect incorrect steps in multimodal reasoning tasks. We hope that our work can inspire more future research and contribute to the development of MLLMs. Our model, data, and benchmark will be released.

Chen, Chao, Guo, Chenghua, Chen, Rufeng, Ma, Guixiang, Zeng, Ming, Liao, Xiangwen, Zhang, Xi, Xie, Sihong

To foster trust in machine learning models, explanations must be faithful and stable for consistent insights. Existing relevant works rely on the ${\ell }_p$ distance for stability assessment, which diverges from human perception. Besides, existing adversarial training (AT) associated with intensive computations may lead to an arms race. To address these challenges, we introduce a novel metric to assess the stability of top-$k$ salient features. We introduce R2ET which trains for stable explanation by efficient and effective regularizer,and analyze R2ET by multi-objective optimization to prove numerical and statistical stability of explanations. Moreover, theoretical connections between R2ET and certified robustness justify R2ET's stability in all attacks. Extensive experiments across various data modalities and model architectures show that R2ET achieves superior stability against stealthy attacks, and generalizes effectively across different explanation methods. The code can be found at https://github.com/ccha005/R2ET.

David A. R. Robin, Killian Bakong, Kevin Scaman

In the context of smooth stochastic optimization with first order methods, we introduce the stability ratio of gradient estimates, as a measure of local relative noise level, from zero for pure noise to one for negligible noise. We show that a schedule-free variant (Stab-SGD) of stochastic gradient descent obtained by just shrinking the learning rate by the stability ratio achieves real adaptivity to noise levels (i.e. without tuning hyperparameters to the gradient's variance), with all key properties of a good schedule-free algorithm: neither plateau nor explosion at intialization, and no saturation of the loss. We believe this theoretical development reveals the importance of estimating the local stability ratio in the construction of well-behaved (last-iterate) schedule-free algorithms, particularly when hyperparameter-tuning budgets are a small fraction of the total budget since noise-adaptivity and cheaper horizon-free tuning are most crucial in this regime.

Xiao Zhang, Xiangyu Han, Xiwen Lai, Yao Sun, Pei Zhang, Xia Liu, Konrad P Kording

Current zero-shot unsupervised segmentation methods based on normalized cuts (NCut) face three key limitations. First, they rely on recursive bipartitions with repeated eigen-decompositions, making them prohibitively expensive at scale. Second, each split requires spectral relaxation followed by rounding, introducing layers of approximation where the final partition may diverge from the true NCut objective. Third, recursive bipartitioning offers no principled assurance of producing a stable $K$-way segmentation, and existing heuristics lack convergence guarantees. We propose \textbf{Falcon}, a proximal-gradient solver that directly optimizes the discrete $K$-way NCut objective without spectral relaxation. We prove linear convergence under the \textit{Kurdyka--\L{}ojasiewicz} (KL) property. Falcon computes closed-form gradient scores weighted by cluster volumes and performs row-wise one-hot proximal updates stabilized by inertia. A monotone backtracking scheme adaptively tunes the proximal parameter, ensuring non-decreasing NCut values. This design preserves discrete feasibility, removes repeated eigen-decomposition, and guarantees convergence. Across six benchmarks, Falcon outperforms the strongest official baseline (DiffCut) by wide margins, e.g., +13.2 mIoU on VOC, +27.7 on COCO-Object, and +3.1 on Cityscapes, while remaining competitive on Pascal Context. It also runs up to an order of magnitude faster than recursive NCut and scales more favorably in memory at high resolution, making it practical for larger token grids. By pairing pretrained foundation models with a principled NCut solver, Falcon sets a new state of the art across six benchmarks and achieves the best performance on 17 of 18 benchmark--encoder pairs, underscoring both its robustness and its generality in bridging the gap between unsupervised and supervised segmentation.

Tu, Zhenfeng, Aranguri Diaz, Santiago Tomas, Jacot, Arthur

The training dynamics of linear networks are well studied in two distinctsetups: the lazy regime and balanced/active regime, depending on theinitialization and width of the network. We provide a surprisinglysimple unifying formula for the evolution of the learned matrix thatcontains as special cases both lazy and balanced regimes but alsoa mixed regime in between the two. In the mixed regime, a part ofthe network is lazy while the other is balanced. More precisely thenetwork is lazy along singular values that are below a certain thresholdand balanced along those that are above the same threshold. At initialization,all singular values are lazy, allowing for the network to align itselfwith the task, so that later in time, when some of the singular valuecross the threshold and become active they will converge rapidly (convergencein the balanced regime is notoriously difficult in the absence ofalignment). The mixed regime is the `best of both worlds': it convergesfrom any random initialization (in contrast to balanced dynamics whichrequire special initialization), and has a low rank bias (absent inthe lazy dynamics). This allows us to prove an almost complete phasediagram of training behavior as a function of the variance at initializationand the width, for a MSE training task.

Michael Harding, Kirthevasan Kandasamy

We consider the problem of identifying the best arm in a multi-armed bandit model. Despite a wealth of literature in the traditional fixed budget and fixed confidence regimes of the best arm identification problem, it still remains a mystery to most practitioners as to how to choose an approach and corresponding budget or confidence parameter. We propose a new formalism to avoid this dilemma altogether by minimizing a risk functional which explicitly balances the performance of the recommended arm and the cost incurred by learning this arm. In this framework, a cost is incurred for each observation during the sampling phase, and upon recommending an arm, a performance penalty is incurred for identifying a suboptimal arm. The learner's goal is to minimize the sum of the penalty and cost. This new regime mirrors the priorities of many practitioners, e.g. maximizing profit in an A/B testing framework, better than classical fixed budget or confidence settings. We derive theoretical lower bounds for the risk of each of two choices for the performance penalty, the probability of misidentification and the simple regret, and propose an algorithm called DBCARE to match these lower bounds up to polylog factors on nearly all problem instances. We then demonstrate the performance of DBCARE on a number of simulated models, comparing to fixed budget and confidence algorithms to show the shortfalls of existing BAI paradigms on this problem.

Wuyang Zhang, Shichao Pei

Retrieval-Augmented Generation (RAG) improves factual grounding in large language models but suffers from substantial latency due to synchronous retrieval. While recent work explores asynchronous retrieval, existing approaches rely on heuristic coordination between retrieval and generation and assume stable information demands during decoding that often break in complex, multi-domain settings. In this paper, we propose an advanced asynchronous retrieval framework that enables predictive prefetching aligned with evolving information needs. The framework explicitly predicts when retrieval should be triggered and what information should be retrieved using three components, a retrieval predictor, a context monitor, and a query generator, by exploiting semantic precursors in generation dynamics that emerge several tokens before uncertainty becomes critical. Experiments on multiple benchmarks demonstrate up to 43.5% end-to-end latency reduction and 62.4% improvement in time-to-first-token, while maintaining answer quality comparable to synchronous RAG baselines.

Goren, Shani, Galil, Ido, El-Yaniv, Ran

Deploying deep neural networks for risk-sensitive tasks necessitates an uncertainty estimation mechanism. This paper introduces hierarchical selective classification, extending selective classification to a hierarchical setting. Our approach leverages the inherent structure of class relationships, enabling models to reduce the specificity of their predictions when faced with uncertainty. In this paper, we first formalize hierarchical risk and coverage, and introduce hierarchical risk-coverage curves. Next, we develop algorithms for hierarchical selective classification (which we refer to as "inference rules"), and propose an efficient algorithm that guarantees a target accuracy constraint with high probability. Lastly, we conduct extensive empirical studies on over a thousand ImageNet classifiers, revealing that training regimes such as CLIP, pretraining on ImageNet21k and knowledge distillation boost hierarchical selective performance.

Haipeng Luo, Spandan Senapati, Vatsal Sharan

Calibration is a fundamental concept that aims at ensuring the reliability of probabilistic predictions by aligning them with real-world outcomes. There is a surge of studies on new calibration measures that are easier to optimize compared to the classical ${\ell }_1$-Calibration while still having strong implications for downstream applications. One recent such example is the work by Fishelson et al. (2025) who show that it is possible to achieve $\tilde{\mathcal{O}}(T^{1/3})$ pseudo ${\ell }_2$-Calibration error via minimizing pseudo swap regret of the squared loss, which in fact implies the same bound for all bounded proper losses with a smooth univariate form. In this work, we significantly generalize their result in the following ways: (a) in addition to smooth univariate forms, our algorithm also simultaneously achieves $\tilde{\mathcal{O}}(T^{1/3})$ swap regret for any proper loss with a twice continuously differentiable univariate form (such as Tsallis entropy); (b) our bounds hold not only for pseudo swap regret that measures losses using the forecaster's distributions on predictions, but also hold for the actual swap regret that measures losses using the forecaster's actual realized predictions. We achieve so by introducing a new stronger notion of calibration called (pseudo) KL-Calibration, which we show is equivalent to the (pseudo) swap regret with respect to log loss. We prove that there exists an algorithm that achieves $\tilde{\mathcal{O}}(T^{1/3})$ KL-Calibration error and provide an explicit algorithm that achieves $\tilde{\mathcal{O}}(T^{1/3})$ pseudo KL-Calibration error. Moreover, we show that the same algorithm achieves $\mathcal{O}(T^{1/3}(\log T{)}^{-\frac{1}{3}}\log (T/\delta ))$ swap regret with probability at least $1-\delta$ for any proper loss with a smooth univariate form, which implies $\tilde{\mathcal{O}}(T^{1/3})$ ${\ell }_2$-Calibration error. A technical contribution of our work is a new randomized rounding procedure and a non-uniform discretization scheme to minimize the swap regret for log loss.

Wulin Xie, YiFan Zhang, Chaoyou Fu, Yang Shi, Jianshu Zeng, Bingyan Nie, Hongkai Chen, Zhang Zhang, Liang Wang

Unified Multimodal Large Language Models (U-MLLMs) have garnered considerable interest for their ability to seamlessly integrate generation and comprehension tasks. However, existing research lacks a unified evaluation standard, often relying on isolated benchmarks to assess these capabilities. Moreover, current work highlights the potential of “mixed-modality generation capabilities” through case studies—such as generating auxiliary lines in images to solve geometric problems, or reasoning through a problem before generating a corresponding image. Despite this, there is no standardized benchmark to assess models on such unified tasks. To address this gap, we introduce MME-Unify, also termed as MME-U, the first open and reproducible benchmark designed to evaluate multimodal comprehension, generation, and mixed-modality generation capabilities. For comprehension and generation tasks, we curate a diverse set of tasks from 12 datasets, aligning their formats and metrics to develop a standardized evaluation framework. For unified tasks, we design five subtasks to rigorously assess how models’ understanding and generation capabilities can mutually enhance each other. Evaluation of 17 U-MLLMs, including Janus-Pro, Bagel, and Gemini2-Flash, reveals significant room for improvement, particularly in areas such as instruction following and image generation quality.

Minh-Tung Luu, Hwanhee Kim, Younghwan Lee, Chang D. Yoo

Conveying complex objectives to reinforcement learning (RL) agents often requires meticulous reward engineering. Preference-based RL (PbRL) offers a promising alternative by learning reward functions from human feedback, but its scalability is hindered by high labeling costs. Inspired by advances in Video Foundation Models (ViFMs), we present Video-based Optimal Transport Preference (VOTP), a semi-supervised framework that learns effective reward functions from only a handful of labels. By leveraging optimal transport to align visual trajectories within the rich representation space of ViFMs, VOTP effectively generates high-fidelity pseudo-labels for large amounts of unlabeled data, substantially reducing human supervision. Extensive experiments across locomotion and manipulation benchmarks demonstrate the superiority of VOTP, which outperforms state-of-the-art offline PbRL methods under limited feedback budgets. We also showcase the robustness of VOTP in the presence of visual distractors and validate its utility on real robotic tasks, where it learns meaningful rewards with minimal human input.

Aditya Sarkar, Yi Li, Jiacheng Cheng, Shlok Mishra, Nuno Vasconcelos

Selective prediction aims to endow predictors with a reject option, to avoid low confidence predictions. However, existing literature has primarily focused on closed-set tasks, such as visual question answering with predefined options or fixed-category classification. This paper considers selective prediction for visual language foundation models, addressing a taxonomy of tasks ranging from closed to open set and from finite to unbounded vocabularies, as in image captioning. We seek training-free approaches of low-complexity, applicable to any foundation model and consider methods based on external vision-language model (VLM) embeddings, like CLIP. This is denoted as $\textit{Plug-and-Play Selective Prediction} (\textbf{\texttt{PaPSP}})$. We identify two key challenges: (1) $\text{instability of the visual-language representations}$, leading to high variance in image-text embeddings, and (2) $\text{poor calibration of similarity scores}$. To address these issues, we propose a $\text{memory augmented}$ $\textbf{\texttt{PaPSP}}$ ($\textbf{\texttt{MA-PaPSP}}$) model, which augments $\textbf{\texttt{PaPSP}}$ with a retrieval dataset of image-text pairs. This is leveraged to reduce embedding variance by averaging retrieved nearest-neighbor pairs and is complemented by the use of contrastive normalization to improve score calibration. Through extensive experiments on multiple datasets, we show that $\textbf{\texttt{MA-PaPSP}}$ outperforms $\textbf{\texttt{PaPSP}}$ and other selective prediction baselines for selective captioning, image-text matching, and fine-grained classification. Source code will be made public.

YANG, CHENYU, Zhu, Xizhou, Zhu, Jinguo, Su, Weijie, Wang, Junjie, Dong, Xuan, Wang, Wenhai, Li, Bin, Zhou, Jie, Qiao, Yu, Dai, Jifeng

Recently, vision model pre-training has evolved from relying on manually annotated datasets to leveraging large-scale, web-crawled image-text data. Despite these advances, there is no pre-training method that effectively exploits the interleaved image-text data, which is very prevalent on the Internet. Inspired by the recent success of compression learning in natural language processing, we propose a novel vision model pre-training method called Latent Compression Learning (LCL) for interleaved image-text data. This method performs latent compression learning by maximizing the mutual information between the inputs and outputs of a causal attention model. The training objective can be decomposed into two basic tasks: 1) contrastive learning between visual representation and preceding context, and 2) generating subsequent text based on visual representation. Our experiments demonstrate that our method not only matches the performance of CLIP on paired pre-training datasets (e.g., LAION), but can also leverage interleaved pre-training data (e.g., MMC4) to learn robust visual representations from scratch, showcasing the potential of vision model pre-training with interleaved image-text data.

Songjun Tu, Jiahao Lin, Qichao Zhang, Xiangyu Tian, Linjing Li, Xiangyuan Lan, Dongbin Zhao

Large reasoning models (LRMs) are proficient at generating explicit, step-by-step reasoning sequences before producing final answers. However, such detailed reasoning can introduce substantial computational overhead and latency, particularly for simple problems. To address this over-thinking problem, we explore how to equip LRMs with adaptive thinking capabilities—enabling them to dynamically decide whether or not to engage in explicit reasoning based on problem complexity. Building on R1-style distilled models, we observe that inserting a simple ellipsis ("...") into the prompt can stochastically trigger either a thinking or no-thinking mode, revealing a latent controllability in the reasoning behavior. Leveraging this property, we propose AutoThink, a multi-stage reinforcement learning (RL) framework that progressively optimizes reasoning policies via stage-wise reward shaping. AutoThink learns to invoke explicit reasoning only when necessary, while defaulting to succinct responses for simpler tasks. Experiments on five mainstream mathematical benchmarks demonstrate that AutoThink achieves favorable accuracy–efficiency trade-offs compared to recent prompting and RL-based pruning methods. It can be seamlessly integrated into any R1-style model, including both distilled and further fine-tuned variants. Notably, AutoThink improves relative accuracy by 6.4\% while reducing token usage by 52\% on DeepSeek-R1-Distill-Qwen-1.5B, establishing a scalable and adaptive reasoning paradigm for LRMs. Project Page: https://github.com/ScienceOne-AI/AutoThink.

Xiaojian Ding, Lin Zhao, Xian Li, Xiaoying Zhu

Incomplete multi-view data, where certain views are entirely missing for some samples, poses significant challenges for traditional multi-view clustering methods. Existing deep incomplete multi-view clustering approaches often rely on static fusion strategies or two-stage pipelines, leading to suboptimal fusion results and error propagation issues. To address these limitations, this paper proposes a novel incomplete multi-view clustering framework based on Hierarchical Semantic Alignment and Cooperative Completion (HSACC). HSACC achieves robust cross-view fusion through a dual-level semantic space design. In the low-level semantic space, consistency alignment is ensured by maximizing mutual information across views. In the high-level semantic space, adaptive view weights are dynamically assigned based on the distributional affinity between individual views and an initial fused representation, followed by weighted fusion to generate a unified global representation. Additionally, HSACC implicitly recovers missing views by projecting aligned latent representations into high-dimensional semantic spaces and jointly optimizes reconstruction and clustering objectives, enabling cooperative learning of completion and clustering. Experimental results demonstrate that HSACC significantly outperforms state-of-the-art methods on five benchmark datasets. Ablation studies validate the effectiveness of the hierarchical alignment and dynamic weighting mechanisms, while parameter analysis confirms the model's robustness to hyperparameter variations. The code is available at \url{https://github.com/XiaojianDing/2025-NeurIPS-HSACC}.

Massil Achab, Emmanuel Bacry, Stéphane Gaı̈ffas, Iacopo Mastromatteo, Jean-François Muzy

We design a new nonparametric method that allows one to estimate the matrix of integrated kernels of a multivariate Hawkes process. This matrix not only encodes the mutual influences of each node of the process, but also disentangles the causality relationships between them. Our approach is the first that leads to an estimation of this matrix

Soham Gadgil, Chris Lin, Su-In Lee

Sparse autoencoders (SAEs) are used to decompose neural network activations into human-interpretable features. Typically, features learned by a single SAE are used for downstream applications. However, it has recently been shown that a single SAE captures only a limited subset of features that can be extracted from the activation space. Motivated by this limitation, we introduce and formalize SAE ensembles. Furthermore, we propose to ensemble multiple SAEs through *naive bagging* and *boosting*. In naive bagging, SAEs trained with different weight initializations are ensembled, whereas in boosting SAEs sequentially trained to minimize the residual error are ensembled. Theoretically, naive bagging and boosting are justified as approaches to reduce reconstruction error. Empirically, we evaluate our ensemble approaches with three settings of language models and SAE architectures. Our empirical results demonstrate that, compared to an expanded SAE that matches the number of features in the ensemble, ensembling SAEs improves the reconstruction of language model activations along with SAE stability. Additionally, on downstream tasks such as concept detection and spurious correlation removal, SAE ensembles achieve better performance, showing improved practical utility.

Mojmir Mutny, Tadeusz Janik, Andreas Krause

A key challenge in science and engineering is to design experiments to learn about some unknown quantity of interest. Classical experimental design optimally allocates the experimental budget into measurements to maximize a notion of utility (e.g., reduction in uncertainty about the unknown quantity). We consider a rich setting, where the experiments are associated with states in a Markov chain, and we can only choose them by selecting a policy controlling the state transitions. This problem captures important applications, from exploration in reinforcement learning to spatial monitoring tasks. We propose an algorithm – markov-design – that efficiently selects policies whose measurement allocation provably converges to the optimal one. The algorithm is sequential in nature, adapting its choice of policies (experiments) using past measurements. In addition to our theoretical analysis, we demonstrate our framework on applications in ecological surveillance and pharmacology.

Wenhao Li, Xianjing Meng, Qiangchang Wang, Zhongyi Han, Zhibin Wu, Yilong Yin

Few-shot learning (FSL) aims to generalize to novel categories with only a few samples. Recent approaches incorporate large language models (LLMs) to enrich visual representations with semantic embeddings derived from class names. However, they overlook progressive and adaptive alignment between vision and language from low-level to high-level semantics, resulting in limited semantic gains. To address these challenges, we propose Dual-level Vision–Language Alignment with Reinforcement Learning gating (DVLA-RL), which consists of Dual-level Semantic Construction (DSC) and RL-gated Attention (RLA). Specifically, DSC conditions LLMs on both class names and support samples to generate discriminative attributes, progressively selects the most relevant ones, and then synthesizes them into coherent class descriptions. This process provides complementary low-level attributes and high-level descriptions, enabling both fine-grained grounding and holistic class understanding. To dynamically integrate dual-level semantics along with the visual network layers, RLA formulates cross-modal fusion as a sequential decision process. A lightweight policy trained with episodic REINFORCE adaptively adjusts the contributions of self-attention and cross-attention to integrate textual and visual tokens. As a result, shallow layers refine local attributes and deep layers emphasize global semantics, enabling more precise cross-modal alignment. This achieves class-specific discrimination and generalized representations with merely a few support samples. DVLA-RL achieves new state-of-the-art performance across nine benchmarks in three diverse FSL scenarios.

Jun Li

As large language models (LLMs) are increasingly composed into heterogeneous multi-agent systems, a fundamental reliability challenge emerges: knowledge and governance **fragment** across agents, leading to composition-dependent behaviors and linear scaling of violations. Two empirical observations motivate our approach: internal correction methods (Reflexion, CRITIC, Self-Refine) fail to enforce execution-time constraints deterministically, and filter-based methods (LlamaGuard) sacrifice task utility without repair. We introduce **Judgment Operators (JO)**, a decision-time framework implementing four-way intervention semantics (*Allow, Edit, Escalate, Deny*) via a portable artifact $J=(\mathcal{C},\mathcal{P})$, enabling minimal repair without modifying agent internals. We establish: (1) **composition-invariant enforcement** with constant violation probability, motivated by a necessity lower bound showing no decentralized scheme achieves $O(1)$ violation probability; (2) **sublinear** mistake accumulation via JO-A; and (3) **semantic preservation** for code transformation. Programmatically verifiable constraints provide the formal foundation; empirically, robustness extends beyond this scope (partial mediation 3--5$\times$ better than theoretical worst-case). JO achieves 0\% observed violation rate (vs. 48--68\% baselines), 13.5--20.5\% zero-shot cross-model transfer where few-shot prompting fails, and provides a **portable, auditable, and composable** interface for governance and capability injection in multi-agent LLM systems.

Pierre Alquier

We tackle the problem of online optimization with a general, possibly unbounded, loss function. It is well known that when the loss is bounded, the exponentially weighted aggregation strategy (EWA) leads to a regret in $\sqrt{T}$ after $T$ steps. In this paper, we study a generalized aggregation strategy, where the weights no longer depend exponentially on the losses. Our strategy is based on Follow The Regularized Leader (FTRL): we minimize the expected losses plus a regularizer, that is here a $\varphi$-divergence. When the regularizer is the Kullback-Leibler divergence, we obtain EWA as a special case. Using alternative divergences enables unbounded losses, at the cost of a worst regret bound in some cases.