Regularization
-
GAPS: Generative Active Pseudo-view Selection for Sparse-View 3D Gaussian Splatting
2026-09-20 08:13:18Hongfei Zhu, Haochen Deng, Sitao Zhang, Ling ZhouAbstract
Novel view synthesis from sparse observations is severely under-constrained. Although 3D Gaussian Splatting (3DGS) enables real-time rendering, it produces floaters, broken geometry, and washed-out backgrounds when trained with few views. We propose an alternating optimization framework that uses a pre-trained image diffusion model to generate geometrically consistent pseudo-views for additional 3DGS supervision. Generation is constrained by depth-conditioned ControlNet, IP-Adapter style transfer, LoRA scene adaptation, and img2img structural anchoring. We introduce Generative Active Pseudo-view Selection (GAPS) to balance reconstruction informativeness and generative reliability when choosing target views. Its annealing schedule shifts from conservative interpolation early in training to exploratory extrapolation later, gradually covering unobserved regions. A dual-criterion admission gate and uncertainty-weighted losses reject unreliable generations, while density-adaptive DropGaussian reduces overfitting in complex scenes. On LLFF with 3/6/9 views, our method improves average PSNR over vanilla 3DGS by 0.40/0.89/0.70 dB. On Mip-NeRF 360 with 12/24 views, the gains are 1.18/0.80 dB. SSIM improves and LPIPS decreases in every setting. Ablations show that active selection and density-adaptive regularization are both necessary; only the full method reduces LPIPS below the no-pseudo-view baseline on unbounded 360-degree scenes.
Abstract (translated)
URL
https://arxiv.org/abs/2609.23436
PDF
-
Energy-based Regularization for Learning Residual Dynamics in Neural MPC for Omnidirectional Aerial Robots
2026-09-19 18:33:47Johannes K\"ubel, Henrik Krauss, Jinjie Li, Moju ZhaoAbstract
Data-driven Model Predictive Control (MPC) has lately been a core research subject in the field of control theory. The combination of an optimal control framework with deep learning paradigms opens up the possibility to accurately tracking control tasks without the need for complex analytical models. However, the system dynamics are often nuanced and the neural model lacks the potential to understand physical properties such as inertia and conservation of energy. In this work, we propose a novel energy-based regularization loss function which is applied to the training of a neural model that learns the residual dynamics of an omnidirectional aerial robot. Our energy-based regularization encourages the neural network to learn stabilizing control corrections. Without the regularization the network misses physical context. The learned residual dynamics are then integrated into the MPC framework. The positional mean absolute error (MAE) are shown to be improved in three real-world experiments by 23% compared to the analytical MPC. We also compare our method to a standard neural MPC implementation without regularization and primarily achieve an increased flight stability implicitly due to smoothening out the acceleration of the robot and thus up to 15% lower MAE. A complimentary video is available at: this https URL. We share our code under: this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2604.14678
PDF
-
CLEAR: Complex Learned Explicit Analytical Regularization for Ultra-Accelerated 4D Flow CMR Reconstruction
2026-09-19 11:06:20German Sh\^ama Wache, Sebastian NeumayerAbstract
While compressed-sensing regularizers enable interpretable reconstruction of 4D Flow CMR through transparent variational objectives, their hand-crafted nature is too restrictive under high acceleration. State-of-the-art learning-based approaches mitigate this, but typically encode regularization implicitly through unrolled network modules, which limits their interpretability. To address this limitation, we propose CLEAR, designed to combine the interpretability of compressed sensing with the flexibility of learned models. To the best of our knowledge, it is the first learned regularizer for a 4D reconstruction task. In the ultra-accelerated \(10\times\)--\(50\times\) regime of the CMRx4DFlow2026 challenge, CLEAR outperforms compressed sensing locally low-rank (LLR) and the popular variational network FlowVN, while using less than 10k parameters and preserving an interpretable regularization structure.
Abstract (translated)
URL
https://arxiv.org/abs/2609.22950
PDF
-
Federated Deep Clustering Networks for High-Dimensional and Heterogeneous Data
2026-09-18 14:29:31Morris Stallmann, Charalampos S. Kouzinopoulos, Marcin Pietrasik, Anna WilbikAbstract
Clustering high-dimensional data is a fundamental task in unsupervised machine learning with applications to a variety of domains. In the centralized data scenario, this task is commonly solved using deep clustering methods that utilize deep neural network architectures to learn clustering-friendly latent space representations. In Federated Learning, where data is distributed between clients and is private, deep clustering methods are less explored. In particular, recently introduced federated deep clustering methods, despite showing very promising performance, still fall short in reliably providing good performance if data across clients are non-identically-independently distributed. In this work, we introduce a generalization of Deep Clustering Networks to the federated scenario, named FedDCN, that simultaneously optimizes a reconstruction loss and a clustering loss. To ensure robustness and latent space alignment in non-identically-independently distributed data scenarios, FedDCN generates synthetic data augmentations, and its learning objective includes a geometric regularization for latent space alignment. Through experimental evaluation, the effectiveness of the approach under IID and non-IID assumptions is demonstrated, and future research directions are identified.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21829
PDF
-
XCalib Depth-Guided Geometric Optimization for Dense Thermal-Visible Video Registration
2026-09-18 13:39:00Aurelien Godet, Gabriel Jobert, Mauro Dalla MuraAbstract
Image registration is a vital preprocessing step in multimodal perception tasks, including image fusion, object detection, and semantic segmentation. In Advanced Driver- Assistance Systems (ADAS), spatial misalignment between visible (RGB) and infrared (IR) cameras -caused by non-coincident optical axes and field-of-view differences- introduces non-uniform parallax and visual ghosting. Classical keypoint-based methods are restricted to global homographies that fail under dynamic depth, while unconstrained dense flow algorithms lack structural regularization and suffer from temporal instability. In this paper, we propose XCalib, an unsupervised dense thermal-visible registration framework that bridges this gap. Rather than serving as an absolute metric calibration tool, XCalib leverages virtual pinhole camera parameterization strictly as a geometric constraint space. By optimizing effective relative pose and intrinsics alongside predicted monocular metric depth, XCalib restricts the search space of spatial displacements to physically valid projection geometries. Our key contributions are: (1) a novel registration paradigm that uses camera parameterization as an implicit regularizer for dense cross-modal warping; (2) Normalized Edges Correlation (NEC), a robust structural similarity metric tailored to cross- spectral alignment; and (3) extensive quantitative and qualitative evaluations across public ADAS datasets, demonstrating superior temporal stability and alignment accuracy over unconstrained dense flow baselines.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21770
PDF
-
Accelerating Dense LLMs via L0-regularized Mixture-of-Experts
2026-09-18 12:08:31Zhenyu Zhang, Jiudong Yang, Zhaowen Tao, Meng ChenAbstract
Large language models (LLMs) achieve strong performance but suffer from slow and costly inference. Existing acceleration methods often lead to noticeable performance degradation, while Mixture-of-Experts (MoE) models require extensive computational resources. In this paper, we propose L0-MoE, a lightweight MoE approach using L0-regularization to accelerate dense LLMs nearly without performance loss. Our method introduces a cluster confusion matrix for domain-aware dataset curation and applies dynamic batching for efficient training. Experiments show that L0-MoE achieves up to 2.5x speedup over dense models while maintaining competitive performance, outperforming existing LLM acceleration baselines.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21672
PDF
-
OneBid: A Unified Auto-Bidding Foundation Model for Diverse oCPX Advertising Scenarios
2026-09-18 09:43:09Yewen Li, Peng Jiang, Yitian Li, Pengfei Lv, Xialong Liu, Peng Jiang, Qingpeng CaiAbstract
Auto-bidding is central to computational advertising, where strategies must maximize advertisers' conversion value under economic constraints. It has evolved from rule-based controllers to reinforcement learning and generative methods such as Decision Transformer (DT). Yet these methods increasingly mismatch the prevailing optimized cost-per-X (oCPX) paradigm, which spans heterogeneous scenarios (e.g., registration, purchase), each served by a separate model, leading to fragmented pipelines and underexploring cross-scenario modeling. Inspired by foundation models like LLMs, unifying these oCPX scenarios into one model raises three challenges: multi-objective control, scalable capacity under strict latency, and safe offline policy improvement. We present OneBid, a unified auto-bidding foundation model that learns a reusable backbone from heterogeneous oCPX logs and adapts it to scenario-specific deployments via offline post-training. Building on DT, OneBid extends single Return-to-Go conditioning to two atomic signals, Return-to-Go for conversion value and Cost-to-Go for cost ratio, plus value-aware regularization on next-action prediction. To absorb distributional heterogeneity, we design a sequence-level Mixture-of-Experts architecture, where shared experts encode cross-scenario knowledge and sparsely-routed experts capture scenario-specific patterns at low latency, yielding consistent scaling with model size and data. During post-training, we align the backbone with scenario preferences via Critic-guided Relative Offline Policy optimization (CROP): a learned critic scores candidate actions group-relatively, avoiding the unsafe online exploration of GRPO-style fine-tuning while constraining policy shift to reduce OOD risk. Validated via online A/B tests and fully deployed at Kuaishou, OneBid delivers an overall +2.2% ADVV gain on oCPX Ads, peaking at +13.1% in the ROAS scenario.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21550
PDF
-
Adaptive Rollout Truncation Based on Epistemic Uncertainty for Efficient Offline World Model Training
2026-09-18 08:32:34Nikodem Sebastian Zymla, Laurin Thiele, Johannes PitzAbstract
Accurate neural world models are central to model-based robotics, where they enable robots to predict future states from previously observed trajectories. Multi-step autoregressive training improves long-horizon prediction, but fixed rollout horizons also increase computational cost and can amplify early training errors when the model is still inaccurate. Existing training schemes typically use the same rollout length throughout optimization, independent of the model's current predictive reliability. We propose an epistemic uncertainty-driven adaptive rollout strategy for offline world model training following an auto-curriculum training scheme. Instead of always unrolling to a fixed horizon, the model terminates autoregressive rollouts once epistemic uncertainty exceeds a threshold calibrated from a warm-up phase. We study two uncertainty estimators: a five-head ensemble with a shared recurrent backbone and Monte Carlo Dropout. A two-stage warm-up procedure stabilizes uncertainty estimates before we enable adaptive truncation. Experiments on ANYmal-D and ANT show that ensemble-based adaptive truncation matches or improves the prediction accuracy of fixed-horizon training and the RWM-U baseline while requiring substantially fewer cumulative rollout steps. Training a world model on ANYmal-D following the presented approach reaches comparable final performance with the baselines with roughly 72% less rollout computation. These results indicate that epistemic uncertainty is useful not only for downstream policy regularization, but also for making world model training itself more compute-efficient.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21482
PDF
-
Anatomically Faithful Artifact Suppression in SENSE Accelerated Brain MRI
2026-09-18 06:26:57Changjing Chai, Bin Huang, Libo Xu, Jian Zhou, Boyang Pan, Kristen W Yeom, Qiyong Gong, Nan-Jie GongAbstract
Background: Four-fold accelerated sensitivity encoding (SENSE4) can shorten brain MRI acquisition time but may amplify noise and result in residual aliasing artifacts after conventional reconstruction. Purpose: To evaluate whether an image-domain refinement framework can improve the quality of SENSE4 brain MRI while preserving anatomical information for quantitative measurements. Methods: In this prospective paired study, 80 participants underwent fully sampled and four-fold accelerated SENSE T1-weighted MRI. We developed an Anatomy-aware Residual Attention Network (ART-Net) to refine accelerated reconstructions through generalized self-attention and correlation-based residual artifact regularization. Participant-level splitting yielded training, validation, and independent test cohort (45/5/30 participants). The independent test cohort underwent quantitative, segmentation-based, and blinded radiologist assessments of image quality and anatomical preservation. Results: ART-Net demonstrated highly competitive reconstruction performance, achieving the highest peak signal-to-noise ratio (31.03 +/- 2.88 dB) and structural similarity index (0.963 +/- 0.022) among evaluated methods. It also demonstrated improved anatomical fidelity, with numerically highest Dice coefficients for medial temporal structures relevant to atrophy assessment (0.8824 +/- 0.0827) and whole-brain regions (0.8857 +/- 0.0885). Moreover, ART-Net improved gradient fidelity, regional contrast preservation, and radiologist-rated structural quality. Conclusion: ART-Net improved agreement between SENSE4 and fully sampled T1-weighted images in a single-center, held-out test cohort while maintaining segmentation-derived anatomical measurements. These findings suggest that ART-Net may support accelerated brain MRI by improving image fidelity and enabling reliable downstream anatomical analysis.
Abstract (translated)
URL
https://arxiv.org/abs/2609.22390
PDF
-
Safe Real-Time Policy Steering via Noise-Space Trajectory Optimization for One-Step Generative Policies
2026-09-18 02:08:12Qingyi Chen, Joseph Ruan, Zachary KingstonAbstract
Generative robot policies can represent diverse, multimodal behaviors, but adapting pretrained policies to deployment-time constraints such as collision avoidance and orientation maintenance remains challenging. Existing inference-time steering methods typically apply gradient guidance through iterative diffusion or flow processes, which can be computationally expensive for real-time control. We propose INSPO, which formulates inference-time steering of one-step generative policies as trajectory optimization in the policy's input noise space. By optimizing the input noise while evaluating constraints on the induced state trajectory, INSPO searches the policy-induced behavior space without directly modifying generated actions. The optimization includes a regularization term that encourages solutions to remain consistent with the policy's input distribution and is solved online using population-based particle optimization. We evaluate INSPO on state- and image-based task-specific policies and generalist vision-language-action policies across Push-T, Can pick-and-place, and LIBERO-Spatial. INSPO improves task success and constraint satisfaction over best-of-N sampling and action projection, while comparing favorably with gradient-guided generation at lower runtime.
Abstract (translated)
URL
https://arxiv.org/abs/2609.21220
PDF
-
Bio-MF: Low-Latency and High-Fidelity EEG-to-fNIRS Cross-Modal Generation for Hybrid Motor-Imagery Brain--Computer Interfaces
2026-09-17 14:52:42Boyuan Zhao, Sifan Zhang, Luping ChenAbstract
Hybrid motor-imagery brain-computer interfaces (MI-BCIs) combining EEG and fNIRS can outperform EEG-only systems by exploiting complementary electrophysiological and hemodynamic information. To obtain such hybrid information when paired EEG-fNIRS acquisition is unavailable or inconvenient, recent studies have focused on EEG-to-fNIRS cross-modal generation. However, existing methods still suffer from slow generation and often require pretraining, limiting their use in real-time MI-BCI scenarios. Although one-step generative models offer an attractive route to low-latency synthesis, removing the iterative refinement process can reduce generation fidelity and introduce non-physiological artifacts. To address these problems, this paper proposes Bio-MF, a latent-free one-step MeanFlow framework for EEG-conditioned fNIRS generation. Bio-MF performs direct signal-space x-prediction, converts this signal-space output into MeanFlow velocity supervision, and completes inference with one network evaluation. To preserve task-relevant hemodynamic structure under heterogeneous sensor layouts, Bio-MF integrates Spatial-Temporal Interactive 4D Encoding, cross-modal classifier-free guidance, and noise-level-gated FFT regularization. On Dataset 1, EEG + synthetic fNIRS improves ACC over EEG-only by 3.37 and 4.15 percentage points for HbR and HbO, respectively. On Dataset 2, the corresponding gains remain 2.98 and 2.50 percentage points under the unseen 64-channel EEG montage. On an RTX PRO 6000 GPU, Bio-MF generates one fNIRS trial in 7.0 ms, corresponding to an 857x speedup over the 1000-step SCDM latency. These results show that Bio-MF enables fast EEG-to-fNIRS synthesis while preserving task-relevant generation quality for downstream hybrid MI decoding. Our code is available at this https URL.
Abstract (translated)
URL
https://arxiv.org/abs/2609.20904
PDF
-
Federated Learning Framework for Privacy-Preserving Kidney Stone Detection
2026-09-17 06:06:03Najiyya Younas, Omar Abdulkader, Yaser Ali Shah, Muhammad Jawad Ikram, Jebran Khan, Amaad KhalilAbstract
Recent innovations in deep learning have significantly enhanced the diagnosis of medical images, although they are based on the use of centralized data storage that pose severe threats to patient privacy and medical data security. To address this issue, this research proposes a Federated Learning (FL) model that is coupled with an optimized YOLOv8 network to detect the kidney stones on a computed tomography (CT) image and at the same time, protect privacy of the patients. The suggested system can help various medical organizations to jointly train a common model without exchanging the information about the patients. This is to ensure that data protection laws like GDPR and HIPAA are adhered to. The residual feature fusion and DropBlock regularization among other architectural improvements are also included in YOLOv8 to enhance detection robustness and minimize overfitting. Experimental analysis carried out on a distributed CT dataset demonstrated that the federated YOLOv8 model has a mAP at 50 of 0.733 and is able to keep the data confidential. Moreover, its lean design facilitates fast edge deployment and real-time inference across a clinical setting. Altogether, these findings indicate that Federated Learning is a safe and efficient solution to AI-assisted diagnosis in contemporary healthcare when combined with the use of sophisticated object detection models.
Abstract (translated)
URL
https://arxiv.org/abs/2609.19740
PDF
-
RGS: Reflection-aware Gaussian Splatting via Learning Geometry Continuity for Reflective Objects
2026-09-16 20:55:24Xiaobiao Du, Yida Wang, Cheng Bi, Kun Zhan, Xin YuAbstract
Gaussian Splatting has significantly improved the quality of novel view synthesis with explicit Gaussian representation. However, we observed that existing 3D Gaussian Splatting methods (3DGS) often suffer from surface collapse issues on reflective regions, and thus produce inferior geometry and low-quality specular. In this work, we propose a physically-based deferred rendering framework, named Reflection-aware Gaussian Splatting (RGS), that can accurately model specular regions and improve novel view synthesis performance. Specifically, we found that a powerful 3D foundation model can provide a strong 3D geometric prior to foster correct geometric modeling. Based on this, we propose a cross-view shape consistency regularization to regularize the geometry surface with the large model prior and cross-view constraints. In this manner, our RGS can produce smoother geometric surfaces on reflective regions while reducing geometric hollows. To further improve rendering results on reflective regions, we present a reflection-aware densification strategy that is designed to capture specular variations across various views. With this strategy, our RGS is able to render novel views of objects in higher quality. Extensive experiments demonstrate our method consistently renders high-quality reflective objects, achieving state-of-the-art performance.
Abstract (translated)
URL
https://arxiv.org/abs/2609.19421
PDF
-
Double descent is the principle of least action
Abstract
The test error of a model plotted against its number of parameters $d$ falls, peaks when the model can just fit the training data, and falls again, exhibiting the double descent phenomenon. We explain the phenomenon with statistical mechanics. The training trajectory of a stochastic gradient-based method is a particle wandering over the energy landscape of the training loss at an induced temperature $T$, and a run that has equilibrated visits every parameter vector of a given training loss equally often, the fundamental postulate of statistical mechanics, with probability given by the Boltzmann distribution. Because training starts at an initial point and has only finite time to diffuse, it carries an effective weight decay, which makes every parameter a quadratic degree of freedom. The equipartition theorem then distributes the energy among the $d$ degrees of freedom in shares of $T/2$, so at a fixed training loss adding parameters lowers the temperature and drives the Boltzmann distribution toward the stationary path. Finally, adding parameters can only lower the $L^2$ norm of the stationary path, so a solution sampled at fixed loss is less likely to be large with increasing $d$, effectively increasing weight regularization.
Abstract (translated)
URL
https://arxiv.org/abs/2609.19076
PDF
-
Learning Where to Focus: Self-Supervised Multi-Scale ViTs for Histopathology
2026-09-16 12:38:41Anabel Stammer, Valay Bundele, Mehran Hosseinzadeh, Hendrik P. A. LenschAbstract
Pathologists diagnose diseases by first locating suspicious tissue and then examining it at higher magnification, whereas self-supervised vision transformers (ViTs) allocate the same spatial resolution to every image region despite diagnostic evidence being sparse and spanning multiple biological scales. Recent pathology foundation models have substantially improved representation quality by scaling training data and model capacity, but largely retain uniform tokenization. We instead investigate whether pathology representations can be improved by learning where to allocate spatial resolution during self-supervised learning. To this end, we propose CRAFT (Coarse-to-fine Region-Adaptive Feature Tokenization), a DINO-based framework that learns image-dependent mixed-scale representations by using self-supervised attention to selectively refine informative regions while preserving coarse context, together with a symmetric cross-scale regularization objective that encourages complementary coarse and fine representations. Across CAMELYON16, TCGA-Lung subtype classification, and TCGA-LUAD survival prediction, CRAFT consistently outperforms comparable-scale self-supervised methods while requiring lower inference computation. Despite using only a compact 22M parameter backbone trained on comparatively small pathology datasets, CRAFT remains competitive with, and often surpasses, substantially larger pathology foundation models.
Abstract (translated)
URL
https://arxiv.org/abs/2609.18578
PDF
-
ActionPiece: Rethinking Action Tokenization for Autoregressive Vision-Language-Action Models
2026-09-16 11:19:54Shijie Lian, Bin Yu, Zhaolong Shen, Xiaopeng Lin, Yichao Du, Zhirui Zhang, Laurence T. Yang, Kai ChenAbstract
Action tokenizers play a central role in autoregressive vision-language-action (VLA) models, determining both the targets for policy training and the executable commands recovered from predicted tokens. Their fidelity is commonly evaluated using pointwise reconstruction metrics such as mean squared error (MSE), yet small individual errors do not fully characterize how faithfully action adjustments across demonstrations are preserved. After compression, similar actions may still cluster around a representative motion, while the adjustments needed for different contexts are diminished, distorted, or even reversed. We introduce physical rank consistency (PRC) to measure how well tokenization preserves local physical distance rankings after reconstruction. Evaluating decoded actions provides a common reference across token vocabularies and decoder architectures, complementing pointwise accuracy with a measure of relational fidelity. We further present ActionPiece, which preserves physical action relationships through joint supervision of representation learning and quantization. Physical rank preservation supervises near-far ordering in encoder and quantized feature distances, while quantization regularization applies the same ordering to codeword assignment distributions. Both objectives augment reconstruction, producing discrete action tokens for standard autoregressive policy learning and execution through a frozen decoder. Under the same Qwen3-VL-4B policy training setup, ActionPiece achieves 94.8% on LIBERO and 68.8% on unseen LIBERO-Plus, with additional evaluations reaching 71.9% on SimplerEnv and 51.5% across VLA-Arena L0-L2. Component ablations show that the two objectives jointly improve PRC and policy success, demonstrating the value of physical relationship supervision for action tokenization.
Abstract (translated)
URL
https://arxiv.org/abs/2609.18487
PDF
-
DR.WILSS: Diffusion-Based Replay for Weakly Supervised Continual Semantic Segmentation
2026-09-16 10:33:51Leon Arthur Marx, Francesco Barbato, Matteo Caligiuri, Pietro ZanuttighAbstract
Weakly supervised class-incremental semantic segmentation (WILSS) aims to train a segmentation model over multiple steps, each introducing new concepts to be learned with only image-level supervision. We introduce this http URL, an innovative approach to address catastrophic forgetting in continual learning using diffusion-based generative replay. Our framework leverages language clues to guide the diffusion process, employing self-inpainting and regularization techniques to efficiently produce replay data, aiding the learning process. By generating high-quality replay data, the information from previously learned classes can be preserved during continual updates, a critical challenge in incremental learning scenarios. To further align the statistics of replay data with those of training samples, we apply LoRAs to the generative model. Experimental results demonstrate state-of-the-art performance across multiple benchmarks and generative architectures, while avoiding storage of training data and the use of additional resource-demanding tools during training. The proposed technique enables an optimal tradeoff between training complexity and inference-time accuracy, making this http URL a promising solution for real-world applications.
Abstract (translated)
URL
https://arxiv.org/abs/2609.18444
PDF
-
WFM: Wiki Foundation Model for Complex Agentic Reasoning
2026-09-16 06:13:33Junnan Dong, Linhao Luo, Senlei Zhang, Gong Chen, Taian Guo, Yifei Yu, Rong Tao, Tao Guo, Qian-Wen Zhang, Siyu An, Ruizhi Qiao, Xing SunAbstract
Real-world agents fundamentally require persistent non-parametric knowledge for dynamic reasoning, i.e., long-term memory and retrieval-augmented generation. While graphs have shown reliable advantages in providing structured evidence, the sparse graph representations naturally restrict machine readability and semantic density required for complex agentic workflows. Driven by this limitation, the entire industry is witnessing a paradigm shift from traditional sparse graphs to LLM Wiki, an agent-native knowledge representation that couples dense document contexts with markdown files containing multi-layered topological linkages. However, parameterizing such rich semantics is challenging to encode dense textual contexts using traditional sparse graph embeddings. Moreover, learning LLM Wiki with existing graph encoders could overwhelm distributed system overheads that hinder deployment in large-scale commercial scenarios. To this end, we propose a novel paradigm Wiki Foundation Model, i.e., WFM, tailored for scalable, agent-native representation and retrieval. Specifically, (i) we formalize a Wiki Graph schema that seamlessly bridges fine-grained structures with dense contexts, maintaining explicit topologies alongside continuous semantics; (ii) A query-conditioned attentive aggregation is tailored for rich wiki message passing and explicit attention variance regularization; (iii) We engineer an infrastructural NCCL boundary exchange protocol that hoists static partition indices and leverages fixed-shape GPU-to-GPU collectives, bypassing CPU serialization and memory copy overheads. Extensive evaluations across five long-term agent memory and multi-hop reasoning benchmarks demonstrate the remarkable performance of WFM, while achieving a 10.5 times training acceleration on distributed clusters.
Abstract (translated)
URL
https://arxiv.org/abs/2609.18182
PDF
-
RAF-VLA: Representation Alignment with the Future for End-to-End Autonomous Driving
2026-09-15 18:39:00Dogun Kim, Yongjae Lee, Joonhee Lim, Yeina Lee, Junhyeok Park, Moogeun Park, Dongsuk KumAbstract
Recent Vision-Language-Action (VLA) models for autonomous driving have incorporated world modeling by predicting future driving scenes alongside driving actions, demonstrating strong planning performance. Future driving scenes are utilized as dense supervision, encouraging the policy to learn rich internal representations useful for planning. However, these World-Modeling VLAs rely on explicit future generation to learn such representations, thereby introducing two key limitations: additional training burden and inference latency. To address these limitations, we propose RAF-VLA (Representation Alignment with the Future), a VLA-based autonomous driving framework that shapes planning-relevant internal representations through direct guidance from future-frame representations. RAF-VLA employs Future-Aligned Supervised Fine-Tuning, in which a straightforward regularization aligns the policy's hidden states with future-frame representations obtained from a pretrained world encoder while learning driving actions. This simple alignment allows RAF-VLA to avoid the training burden and inference latency associated with future generation. Extensive experiments on the NAVSIM benchmark show that RAF-VLA achieves competitive planning performance against state-of-the-art VLA planners with substantially fewer training samples seen. Moreover, RAF-VLA incurs only 3.8% training overhead and a negligible 1 ms inference overhead.
Abstract (translated)
URL
https://arxiv.org/abs/2609.17728
PDF
-
Dissecting Motion-Prior Regularization for Data-Scarce Robotic Insertion
Abstract
This study asks whether training-time motion-prior regularization can improve insertion success when a diffusion policy is learned from only 15 demonstrations. Minimum jerk discourages abrupt changes in predicted translational acceleration; speed-curvature regularization instead couples movement speed to path geometry. These are candidate mechanisms for task completion, not safety guarantees. We compare the priors individually and jointly, neither prior, and generic smoothness, with 80 real-robot trials per setting pooled over four recorded condition classes. Joint and minimum-jerk-only settings each achieved 70/80 successes (87.5%), versus 69/80 (86.3%) for speed-curvature only, 66/80 (82.5%) for neither prior, and 67/80 (83.8%) for generic smoothness. Success rates and Wilson 95% confidence intervals are visualized for direct comparison. Joint regularization exceeded neither by 5.0 percentage points but provided no observed gain over minimum jerk alone. The results motivate minimum jerk as the simpler candidate for replication, without establishing synergy, biomechanical specificity, improved safety, or distribution-shift robustness.
Abstract (translated)
URL
https://arxiv.org/abs/2609.17484
PDF
