This TV-regularized convex program on the space of measures allows to recover a sparse measure using a noisy observation from an appropriate measurement operator. While previous works have uncovered the central role played by this operator and its associated kernel in order to get estimation error bounds, the latter requires a technical local positive curvature (LPC) assumption to be verified on a case-by-case basis. In practice, this yields only few LPC-kernels for which this condition is proved. At the heart of our contribution lies the kernel switch, which uncouples the model kernel from the LPC assumption: it enables to leverage any known LPC-kernel as a pivot kernel to prove error bounds, provided embedding conditions are verified between the model and pivot RKHS. We increment the list of LPC-kernels, proving that the "sinc-4" kernel, used for signal recovery and mixture problems, does satisfy the LPC assumption. Furthermore, we also show that the BLASSO localisation error around the true support decreases with the noise level, leading to effective near regions. This improves on known results where this error is fixed with some parameters depending on the model kernel. We illustrate the interest of our results in the case of translation-invariant mixture model estimation, using bandlimiting smoothing and sketching techniques to reduce the computational burden of BLASSO.

Undesirable biases encoded in the data are key drivers of algorithmic discrimination. Their importance is widely recognized in the algorithmic fairness literature, as well as legislation and standards on anti-discrimination in AI. Despite this recognition, data biases remain understudied, hindering the development of computational best practices for their detection and mitigation. In this work, we present three common data biases and study their individual and joint effect on algorithmic discrimination across a variety of datasets, models, and fairness measures. We find that underrepresentation of vulnerable populations in training sets is less conducive to discrimination than conventionally affirmed, while combinations of proxies and label bias can be far more critical. Consequently, we develop dedicated mechanisms to detect specific types of bias, and combine them into a preliminary construct we refer to as the Data Bias Profile (DBP). This initial formulation serves as a proof of concept for how different bias signals can be systematically documented. Through a case study with popular fairness datasets, we demonstrate the effectiveness of the DBP in predicting the risk of discriminatory outcomes and the utility of fairness-enhancing interventions. Overall, this article bridges algorithmic fairness research and anti-discrimination policy through a data-centric lens.

Batch normalization (BN) is a ubiquitous operation in deep neural networks used primarily to achieve stability and regularization during network training. BN involves feature map centering and scaling using sample means and variances, respectively. Since these statistics are being estimated across the feature maps within a batch, this problem is ideally suited for the application of Stein's shrinkage estimation, which leads to a better, in the mean-squared-error sense, estimate of the mean and variance of the batch. In this paper, we prove that the Stein shrinkage estimator for the mean and variance dominates over the sample mean and variance estimators in the presence of adversarial attacks when modeling these attacks using sub-Gaussian distributions. This facilitates and justifies the application of Stein shrinkage to estimate the mean and variance parameters in BN and use it in image classification (segmentation) tasks with and without adversarial attacks. We present SOTA performance results using this Stein corrected batch norm in a standard ResNet architecture applied to the task of image classification using CIFAR-10 data, 3D CNN on PPMI (neuroimaging) data and image segmentation using HRNet on Cityscape data with and without adversarial attacks.

Data depths are score functions that quantify in an unsupervised fashion how central is a point inside a distribution, with numerous applications such as anomaly detection, multivariate or functional data analysis, arising across various fields. The halfspace depth was the first depth to aim at generalising the notion of quantile beyond the univariate case. Among the existing variety of depth definitions, it remains one of the most used notions of data depth. Taking a different angle from the quantile point of view, we show that the halfspace depth can also be regarded as the minimum loss of a set of classifiers for a specific labelling of the points. By changing the loss or the set of classifiers considered, this new angle naturally leads to a family of "loss depths", extending to well-studied classifiers such as, e.g., SVM or logistic regression, among others. This framework directly inherits computational efficiency of existing machine learning algorithms as well as their fast statistical convergence rates, and opens the data depth realm to the high-dimensional setting. Furthermore, the new loss depths highlight a connection between the dataset and the right amount of complexity or simplicity of the classifiers. The simplicity of classifiers as well as the interpretation as a risk makes our new kind of data depth easy to explain, yet efficient for anomaly detection, as is shown by experiments.

We propose a concept-based adversarial attack framework that extends beyond single-image perturbations by adopting a probabilistic perspective. Rather than modifying a single image, our method operates on an entire concept -- represented by a probabilistic generative model or a set of images -- to generate diverse adversarial examples. Preserving the concept is essential, as it ensures that the resulting adversarial images remain identifiable as instances of the original underlying category or identity. By sampling from this concept-based adversarial distribution, we generate images that maintain the original concept but vary in pose, viewpoint, or background, thereby misleading the classifier. Mathematically, this framework remains consistent with traditional adversarial attacks in a principled manner. Our theoretical and empirical results demonstrate that concept-based adversarial attacks yield more diverse adversarial examples and effectively preserve the underlying concept, while achieving higher attack efficiency.

In designing generative models, it is commonly believed that in order to learn useful latent structure, we face a fundamental tension between expressivity and structure. In this paper we challenge this view by proposing a new approach to training arbitrarily expressive generative models that simultaneously learn disentangled latent structure. This is accomplished by adding a simple decoder-only module to the head of an existing decoder block that can be arbitrarily complex. The module learns to process concept information by implicitly inverting linear representations from an encoder. Inspired by the notion of intervention in causal graphical models, our module selectively modifies its architecture during training, allowing it to learn a compact joint model over different contexts. We show how adding this module leads to disentangled representations that can be composed for out-of-distribution generation. To further validate our proposed approach, we prove a new identifiability result that extends existing work on identifying structured representations in nonlinear models.

--We introduce DeST A2.5-Audio, a general-purpose Large Audio Language Model (LALM) designed for robust auditory perception and instruction-following, without requiring task-specific audio instruction-tuning. Recent LALMs typically augment Large Language Models (LLMs) with auditory capabilities by training on large-scale, manually curated or LLM-synthesized audio-instruction datasets. However, these approaches have often suffered from the catastrophic forgetting of the LLM's original language abilities. T o address this, we revisit the data construction pipeline and propose DeST A, a self-generated cross-modal alignment strategy in which the backbone LLM generates its own training targets. This approach preserves the LLM's native language proficiency while establishing effective audio-text alignment, thereby enabling zero-shot generalization without task-specific tuning. Using DeST A, we construct DeST A-AQA5M, a large-scale, task-agnostic dataset containing 5 million training samples derived from 7,000 hours of audio spanning 50 diverse datasets, including speech, environmental sounds, and music. DeST A2.5-Audio achieves state-of-the-art or competitive performance across a wide range of audio-language benchmarks, including Dynamic-SUPERB, MMAU, SAKURA, Speech-IFEval, and V oiceBench. Comprehensive comparative studies demonstrate that our self-generated strategy outperforms widely adopted data construction and training strategies in both auditory perception and instruction-following capabilities. Our findings underscore the importance of carefully designed data construction in LALM development and offer practical insights for building robust, general-purpose LALMs. HE development of general-purpose artificial intelligence has become a central focus in contemporary AI research, driven by the remarkable performance of large language models (LLMs) across various natural language understanding and generation tasks [1]-[7]. Building on these advancements, a promising direction is to equip LLMs with multi-modal understanding capabilities, leading to the emergence of Large Audio Language Models (LALMs) [8]-[22] and Large Vision Language Models (L VLMs) [23]-[27]. This paper focuses on building a general-purpose LALM, illustrated in Figure 1. To develop a general-purpose LALM, two core capabilities are essential: auditory perception and instruction-following. Auditory perception refers to the comprehensive processing of auditory information, including speech, non-verbal cues, background sounds, and music.

Estimating physical parameters or material properties from experimental observations is a common objective in many areas of physics and material science. In many experiments, especially in shock physics, radiography is the primary means of observing the system of interest. However, radiography does not provide direct access to key state variables, such as density, which prevents the application of traditional parameter estimation approaches. Here we focus on flyer plate impact experiments on porous materials, and resolving the underlying parameterized equation of state (EoS) and crush porosity model parameters given radiographic observation(s). We use machine learning as a tool to demonstrate with high confidence that using only high impact velocity data does not provide sufficient information to accurately infer both EoS and crush model parameters, even with fully resolved density fields or a dynamic sequence of images. We thus propose an observable data set consisting of low and high impact velocity experiments/simulations that capture different regimes of compaction and shock propagation, and proceed to introduce a generative machine learning approach which produces a posterior distribution of physical parameters directly from radiographs. We demonstrate the effectiveness of the approach in estimating parameters from simulated flyer plate impact experiments, and show that the obtained estimates of EoS and crush model parameters can then be used in hydrodynamic simulations to obtain accurate and physically admissible density reconstructions. Finally, we examine the robustness of the approach to model mismatches, and find that the learned approach can provide useful parameter estimates in the presence of out-of-distribution radiographic noise and previously unseen physics, thereby promoting a potential breakthrough in estimating material properties from experimental radiographic images.

The appearance of singularities in the function of interest constitutes a fundamental challenge in scientific computing. It can significantly undermine the effectiveness of numerical schemes for function approximation, numerical integration, and the solution of partial differential equations (PDEs), etc. The problem becomes more sophisticated if the location of the singularity is unknown, which is often encountered in solving PDEs. Detecting the singularity is therefore critical for developing efficient adaptive methods to reduce computational costs in various applications. In this paper, we consider singularity detection in a purely data-driven setting. Namely, the input only contains given data, such as the vertex set from a mesh. To overcome the limitation of the raw unlabeled data, we propose a self-supervised learning (SSL) framework for estimating the location of the singularity. A key component is a filtering procedure as the pretext task in SSL, where two filtering methods are presented, based on $k$ nearest neighbors and kernel density estimation, respectively. We provide numerical examples to illustrate the potential pathological or inaccurate results due to the use of raw data without filtering. Various experiments are presented to demonstrate the ability of the proposed approach to deal with input perturbation, label corruption, and different kinds of singularities such interior circle, boundary layer, concentric semicircles, etc.

High-stakes decision-making involves navigating multiple competing objectives with expensive evaluations. For instance, in brachytherapy, clinicians must balance maximizing tumor coverage (e.g., an aspirational target or soft bound of >95% coverage) against strict organ dose limits (e.g., a non-negotiable hard bound of <601 cGy to the bladder), with each plan evaluation being resource-intensive. Selecting Pareto-optimal solutions that match implicit preferences is challenging, as exhaustive Pareto frontier exploration is computationally and cognitively prohibitive, necessitating interactive frameworks to guide users. While decision-makers (DMs) often possess domain knowledge to narrow the search via such soft-hard bounds, current methods often lack systematic approaches to iteratively refine these multi-faceted preference structures. Critically, DMs must trust their final decision, confident they haven't missed superior alternatives; this trust is paramount in high-consequence scenarios. We present Active-MoSH, an interactive local-global framework designed for this process. Its local component integrates soft-hard bounds with probabilistic preference learning, maintaining distributions over DM preferences and bounds for adaptive Pareto subset refinement. This is guided by an active sampling strategy optimizing exploration-exploitation while minimizing cognitive burden. To build DM trust, Active-MoSH's global component, T-MoSH, leverages multi-objective sensitivity analysis to identify potentially overlooked, high-value points beyond immediate feedback. We demonstrate Active-MoSH's performance benefits through diverse synthetic and real-world applications. A user study on AI-generated image selection further validates our hypotheses regarding the framework's ability to improve convergence, enhance DM trust, and provide expressive preference articulation, enabling more effective DMs.