Magnetic Resonance Imaging (MRI) is the primary imaging modality used in the diagnosis, assessment, and treatment planning for brain pathologies. However, most automated MRI analysis tools, such as segmentation and registration pipelines, are optimized for healthy anatomies and often fail when confronted with large lesions such as tumors. To overcome this, image inpainting techniques aim to locally synthesize healthy brain tissues in tumor regions, enabling the reliable application of general-purpose tools. In this work, we systematically evaluate state-of-the-art inpainting models and observe a saturation in their standalone performance. In response, we introduce a methodology combining model ensembling with efficient post-processing strategies such as median filtering, histogram matching, and pixel averaging. Further anatomical refinement is achieved via a lightweight U-Net enhancement stage. Comprehensive evaluation demonstrates that our proposed pipeline improves the anatomical plausibility and visual fidelity of inpainted regions, yielding higher accuracy and more robust outcomes than individual baseline models. By combining established models with targeted post-processing, we achieve improved and more accessible in-painting outcomes, supporting broader clinical deployment and sustainable, resource-conscious research.

Learning to Rank (LTR) methods generally assume that each document in a top-K ranking is presented in an equal format. However, previous work has shown that users' perceptions of relevance can be changed by varying presentations, i.e., allocating more vertical space to some documents to provide additional textual or image information. Furthermore, presentation length can also redirect attention, as users are more likely to notice longer presentations when scrolling through results. Deciding on the document presentation lengths in a fixed vertical space ranking is an important problem that has not been addressed by existing LTR methods. We address this gap by introducing the variable presentation length ranking task, where simultaneously the ordering of documents and their presentation length is decided. Despite being a generalization of standard ranking, we show that this setting brings significant new challenges: Firstly, the probability ranking principle no longer applies to this setting, and secondly, the problem cannot be divided into separate ordering and length selection tasks. We therefore propose VLPL - a new family of Plackett-Luce list-wise gradient estimation methods for the joint optimization of document ordering and lengths. Our semi-synthetic experiments show that VLPL can effectively balance the expected exposure and attractiveness of all documents, achieving the best performance across different ranking settings. Furthermore, we observe that even simple length-aware methods can achieve significant performance improvements over fixed-length models. Altogether, our theoretical and empirical results highlight the importance and difficulties of combining document presentation with LTR.

We present a customizable soft haptic system that integrates modular hardware with an information-theoretic algorithm to personalize feedback for different users and tasks. Our platform features modular, multi-degree-of-freedom pneumatic displays, where different signal types, such as pressure, frequency, and contact area, can be activated or combined using fluidic logic circuits. These circuits simplify control by reducing reliance on specialized electronics and enabling coordinated actuation of multiple haptic elements through a compact set of inputs. Our approach allows rapid reconfiguration of haptic signal rendering through hardware-level logic switching without rewriting code. Personalization of the haptic interface is achieved through the combination of modular hardware and software-driven signal selection. To determine which display configurations will be most effective, we model haptic communication as a signal transmission problem, where an agent must convey latent information to the user. We formulate the optimization problem to identify the haptic hardware setup that maximizes the information transfer between the intended message and the user's interpretation, accounting for individual differences in sensitivity, preferences, and perceptual salience. We evaluate this framework through user studies where participants interact with reconfigurable displays under different signal combinations. Our findings support the role of modularity and personalization in creating multimodal haptic interfaces and advance the development of reconfigurable systems that adapt with users in dynamic human-machine interaction contexts.

When a deep ReLU network is initialized with small weights, GD is at first dominated by the saddle at the origin in parameter space. We study the so-called escape directions, which play a similar role as the eigenvectors of the Hessian for strict saddles. We show that the optimal escape direction features a low-rank bias in its deeper layers: the first singular value of the $\ell$-th layer weight matrix is at least $\ell^{\frac{1}{4}}$ larger than any other singular value. We also prove a number of related results about these escape directions. We argue that this result is a first step in proving Saddle-to-Saddle dynamics in deep ReLU networks, where GD visits a sequence of saddles with increasing bottleneck rank.

Reward models (RMs) play a pivotal role in aligning large language models (LLMs) with human values. However, noisy preferences in human feedback can lead to reward misgeneralization - a phenomenon where reward models learn spurious correlations or overfit to noisy preferences, which poses important challenges to the generalization of RMs. This paper systematically analyzes the characteristics of preference pairs and aims to identify how noisy preferences differ from human-aligned preferences in reward modeling. Our analysis reveals that noisy preferences are difficult for RMs to fit, as they cause sharp training fluctuations and irregular gradient updates. These distinctive dynamics suggest the feasibility of identifying and excluding such noisy preferences. Empirical studies demonstrate that policy LLM optimized with a reward model trained on the full preference dataset, which includes substantial noise, performs worse than the one trained on a subset of exclusively high quality preferences. To address this challenge, we propose an online Collaborative Reward Modeling (CRM) framework to achieve robust preference learning through peer review and curriculum learning. In particular, CRM maintains two RMs that collaboratively filter potential noisy preferences by peer-reviewing each other's data selections. Curriculum learning synchronizes the capabilities of two models, mitigating excessive disparities to promote the utility of peer review. Extensive experiments demonstrate that CRM significantly enhances RM generalization, with up to 9.94 points improvement on RewardBench under an extreme 40\% noise. Moreover, CRM can seamlessly extend to implicit-reward alignment methods, offering a robust and versatile alignment strategy.

Figure 1: Given a few input images of a real world product, our system can generate images that not only maintain high fidelity to the original product, but also recontextualize it in novel settings beyond background changes: from showcasing it in a new perspective, adding object occlusions, to creating different and realistic lighting conditions. We present a framework for high-fidelity product image recontextualization using text-to-image diffusion models and a novel data augmentation pipeline. This pipeline leverages image-to-video diffusion, in/outpainting & negatives to create synthetic training data, addressing limitations of real-world data collection for this task. Our method improves the quality and diversity of generated images by disentangling product representations and enhancing the model's understanding of product characteristics. Evaluation on the ABO dataset and a private product dataset, using automated metrics and human assessment, demonstrates the effectiveness of our framework in generating realistic and compelling product visualizations, with implications for applications such as e-commerce and virtual product showcasing.

Based on a new atomic norm, we propose a new convex formulation for sparse matrix factorization problems in which the number of non-zero elements of the factors is assumed fixed and known. The formulation counts sparse PCA with multiple factors, subspace clustering and low-rank sparse bilinear regression as potential applications.

Distributed learning algorithms, such as the ones employed in Federated Learning (FL), require communication compression to reduce the cost of client uploads. The compression methods used in practice are often biased, which require error feedback to achieve convergence when the compression is aggressive. In turn, error feedback requires client-specific control variates, which directly contradicts privacy-preserving principles and requires stateful clients. In this paper, we propose Compressed Aggregate Feedback (CAFe), a novel distributed learning framework that allows highly compressible client updates by exploiting past aggregated updates, and does not require control variates. We consider Distributed Gradient Descent (DGD) as a representative algorithm and provide a theoretical proof of CAFe's superiority to Distributed Compressed Gradient Descent (DCGD) with biased compression in the non-smooth regime with bounded gradient dissimilarity. Experimental results confirm that CAFe consistently outperforms distributed learning with direct compression and highlight the compressibility of the client updates with CAFe.