Principal Component Analysis is a key technique for reducing the complexity of high-dimensional data while preserving its fundamental data structure, ensuring models remain stable and interpretable. This is achieved by transforming the original variables into a new set of uncorrelated variables (principal components) based on the covariance structure of the original variables. However, since the traditional maximum likelihood covariance estimator does not accurately converge to the true covariance matrix, the standard principal component analysis performs poorly as a dimensionality reduction technique in high-dimensional scenarios $n

Abstract--Real-world machine learning applications are often hindered by two critical challenges: distribution shift and label noise. Networks inherently tend to overfit to redundant, uninformative features present in the training distribution, which undermines their ability to generalize effectively to the target domain's distribution. The presence of noisy data further exacerbates this issue by inducing additional overfitting to noise, causing existing domain generalization methods to fail in effectively distinguishing invariant features from spurious ones. We also introduce a weighted loss function that dynamically adjusts the contribution of each sample based on its distance to the corresponding NLP anchor, thereby improving the model's resilience to noisy labels. Generalization (DG) has emerged as a pivotal algorithm in machine learning, aiming to develop models that can maintain high performance on previously unseen environments--or domains. T raditional methods often assume that training and test data share the same distribution, yet in real-world scenarios, there is frequently a substantial shift between these distributions. This phenomenon, widely referred to as domain shift, can cause severe performance degradation in tasks spanning computer vision, natural language processing, and medical image analysis [1]. As shown in Figure 1(a)(b), even within the same class label, the distribution of feature representations can vary considerably . This variation may stem from differences in image acquisition conditions--such as lighting variations, changes in pose, or complex background environments--and even from more subtle domain-specific factors like sensor noise or camera calibration differences. Such intra-class variability poses a significant challenge for developing accurate and adaptable models, which must learn to extract invariant features that capture the true semantic essence of the class while ignoring irrelevant variations. Lin, Z. Zhang is with Worcester Polytechnic Institute, Worcester, MA, 01890. L.Wang is with Carnegie Mellon University, Pittsburgh, P A, 15213. Y .Wang is with Peking University, Beijing, China, 100871. Z.Li, W.Lu is with T singhua University, Beijing, China, 100190. K.Y amada is with T ohoku University, Sendai, Japan, 980-8572.

The behavior of multivariate dynamical processes is often governed by underlying structural connections that relate the components of the system. For example, brain activity which is often measured via time series is determined by an underlying structural graph, where nodes represent neurons or brain regions and edges represent cortical connectivity. Existing methods for inferring structural connections from observed dynamics, such as correlation-based or spectral techniques, may fail to fully capture complex relationships in high-dimensional time series in an interpretable way. Here, we propose the use of path signatures--a mathematical framework that encodes geometric and temporal properties of continuous paths--to address this problem. Path signatures provide a reparametrization-invariant characterization of dynamical data and, in particular, can be used to compute the lead matrix which reveals lead-lag phenomena. We showcase our approach on time series from coupled oscillators in the Kuramoto model defined on a stochastic block model graph, termed the Kuramoto stochastic block model (KSBM). Using mean-field theory and Gaussian approximations, we analytically derive reduced models of KSBM dynamics in different temporal regimes and theoretically characterize the lead matrix in these settings. Leveraging these insights, we propose a novel signature-based community detection algorithm, achieving exact recovery of structural communities from observed time series in multiple KSBM instances. Our results demonstrate that path signatures provide a novel perspective on analyzing complex neural data and other high-dimensional systems, explicitly exploiting temporal functional relationships to infer underlying structure.

Prediction tasks in digital pathology are challenging due to the massive size of whole-slide images (WSIs) and the weak nature of training signals. Advances in computing, data availability, and self-supervised learning (SSL) have paved the way for slide-level foundation models (SLFMs) that can improve prediction tasks in low-data regimes. However, working with these models is challenging, with issues such as catastrophic forgetting during fine-tuning and under-utilization of shared information between tasks and modalities. To overcome these two challenges, we propose ModalTune, a novel fine-tuning framework which introduces the Modal Adapter to integrate new modalities without modifying SLFM weights. Additionally, we use large-language models (LLMs) to encode labels as text, capturing semantic relationships and enhancing generalization across multiple tasks and cancer types in a single training recipe. ModalTune achieves state-of-the-art (SOTA) results against both uni-modal and multi-modal models across four cancer types, jointly improving survival and cancer subtype prediction while remaining competitive in pan-cancer settings. Additionally, we show ModalTune is highly generalizable to two out-of-distribution (OOD) datasets. To our knowledge, this is the first unified fine-tuning framework for multi-modal, multi-task, and pan-cancer modeling in digital pathology.

The dual challenges of prohibitive communication overhead and the impracticality of gradient computation due to data privacy or black-box constraints in distributed systems motivate this work on communication-constrained gradient-free optimization. We propose a stochastic distributed zeroth-order algorithm (Com-DSZO) requiring only two function evaluations per iteration, integrated with general compression operators. Rigorous analysis establishes its sublinear convergence rate for both smooth and nonsmooth objectives, while explicitly elucidating the compression-convergence trade-off. Furthermore, we develop a variance-reduced variant (VR-Com-DSZO) under stochastic mini-batch feedback. The empirical algorithm performance are illustrated with numerical examples.

Human-robot collaboration (HRC) refers to scenarios Various psychophysiological-based methods have in which humans and robots work collaboratively toward a been employed to interpret psychological phenomena within common goal, sharing tasks and responsibilities in a way the context of HRC by measuring the brain and physiological that capitalizes on the strengths of both parties [3]. As activity of workers, such as electroencephalography construction tasks become increasingly complex and timesensitive, (EEG) for brain activity [73], photoplethysmography (PPG), the integration of collaborative robots, or cobots, electrocardiography (ECG) for cardiac activity [7], and into the construction industry has emerged as a solution to electrodermal activity (EDA) for skin response [8]. Given all enhance efficiency and simultaneously mitigate operational the merits of these technologies, some initial endeavors on risks [86, 90]. However, real-world deployment of HRC psychophysiological methods for HRC in construction have in construction confronts multifaceted challenges, such as been made. For instance, real-time feedback from individual's trust in robotic capabilities [21], frequent reconfigurations physiological responses [21] and cognitive load [50] of working conditions [43], and communication in noisy has been used to allow cobots to adjust their behavior (e.g., and unstructured environments [24]. These challenges are accelerate, stop, slow down) in response to the changing exacerbated by the reliability and safety issues inherent in workers' conditions. However, studies on wearable-based complicated and dynamic construction activities and environments psychophysiological methods for the construction industry (e.g., human dynamics, non-deterministic features, to date are still limited and embryonic, primarily focusing and the presence of various materials) [49, 50]. To address on interpreting a specific dimension of worker status. While these limitations, the development of HRC is shifting these methods hold promise for advancing human-centric from performance-oriented approaches to human-centrality robot collaboration in construction, their potential has not yet paradigms, emphasizing a comprehensive interpretation of been fully explored, and current applications remain largely collaborative behaviors between humans and their robot experimental.

Transaction graphs, which represent financial and trade transactions between entities such as bank accounts and companies, can reveal patterns indicative of financial crimes like money laundering and fraud. However, effective detection of such cases requires node and edge classification methods capable of addressing the unique challenges of transaction graphs, including rich edge features, multigraph structures and temporal dynamics. To tackle these challenges, we propose TeMP-TraG, a novel graph neural network mechanism that incorporates temporal dynamics into message passing. TeMP-TraG prioritises more recent transactions when aggregating node messages, enabling better detection of time-sensitive patterns. We demonstrate that TeMP-TraG improves four state-of-the-art graph neural networks by 6.19% on average. Our results highlight TeMP-TraG as an advancement in leveraging transaction graphs to combat financial crime.

Point cloud (PC) processing tasks-such as completion, upsampling, denoising, and colorization-are crucial in applications like autonomous driving and 3D reconstruction. Despite substantial advancements, prior approaches often address each of these tasks independently, with separate models focused on individual issues. However, this isolated approach fails to account for the fact that defects like incompleteness, low resolution, noise, and lack of color frequently coexist, with each defect influencing and correlating with the others. Simply applying these models sequentially can lead to error accumulation from each model, along with increased computational costs. To address these challenges, we introduce SuperPC, the first unified diffusion model capable of concurrently handling all four tasks. Our approach employs a three-level-conditioned diffusion framework, enhanced by a novel spatial-mix-fusion strategy, to leverage the correlations among these four defects for simultaneous, efficient processing. We show that SuperPC outperforms the state-of-the-art specialized models as well as their combination on all four individual tasks.

With the availability of virtually infinite number text documents in digital format, automatic comparison of textual data is essential for extracting meaningful insights that are difficult to identify manually. Many existing tools, including AI and large language models, struggle to provide precise and explainable insights into textual similarities. In many cases they determine the similarity between documents as reflected by the text, rather than the similarities between the subjects being discussed in these documents. This study addresses these limitations by developing an n-gram analysis framework designed to compare documents automatically and uncover explainable similarities. A scoring formula is applied to assigns each of the n-grams with a weight, where the weight is higher when the n-grams are more frequent in both documents, but is penalized when the n-grams are more frequent in the English language. Visualization tools like word clouds enhance the representation of these patterns, providing clearer insights. The findings demonstrate that this framework effectively uncovers similarities between text documents, offering explainable insights that are often difficult to identify manually. This non-parametric approach provides a deterministic solution for identifying similarities across various fields, including biographies, scientific literature, historical texts, and more. Code for the method is publicly available.

Deep learning systems are optimized for clusters with homogeneous resources. However, heterogeneity is prevalent in computing infrastructure across edge, cloud and HPC. When training neural networks using stochastic gradient descent techniques on heterogeneous resources, performance degrades due to stragglers and stale updates. In this work, we develop an adaptive batch-scaling framework called OmniLearn to mitigate the effects of heterogeneity in distributed training. Our approach is inspired by proportional controllers to balance computation across heterogeneous servers, and works under varying resource availability. By dynamically adjusting worker mini-batches at runtime, OmniLearn reduces training time by 14-85%. We also investigate asynchronous training, where our techniques improve accuracy by up to 6.9%.