Sensory Temporal Action Detection (STAD) aims to localize and classify human actions within long, untrimmed sequences captured by non-visual sensors such as WiFi or inertial measurement units (IMUs). Unlike video-based TAD, STAD poses unique challenges due to the low-dimensional, noisy, and heterogeneous nature of sensory data, as well as the real-time and resource constraints on edge devices. While recent STAD models have improved detection performance, their high computational cost hampers practical deployment. In this paper, we propose SlimSTAD, a simple yet effective framework that achieves both high accuracy and low latency for STAD. SlimSTAD features a novel Decoupled Channel Modeling (DCM) encoder, which preserves modality-specific temporal features and enables efficient inter-channel aggregation via lightweight graph attention. An anchor-free cascade predictor then refines action boundaries and class predictions in a two-stage design without dense proposals. Experiments on two real-world datasets demonstrate that SlimSTAD outperforms strong video-derived and sensory baselines by an average of 2.1 mAP, while significantly reducing GFLOPs, parameters, and latency, validating its effectiveness for real-world, edge-aware STAD deployment.

Machine fault diagnosis (FD) is a critical task for predictive maintenance, enabling early fault detection and preventing unexpected failures. Despite its importance, existing FD models are operation-specific with limited generalization across diverse datasets. Foundation models (FM) have demonstrated remarkable potential in both visual and language domains, achieving impressive generalization capabilities even with minimal data through few-shot or zero-shot learning. However, translating these advances to FD presents unique hurdles. Unlike the large-scale, cohesive datasets available for images and text, FD datasets are typically smaller and more heterogeneous, with significant variations in sampling frequencies and the number of channels across different systems and applications. This heterogeneity complicates the design of a universal architecture capable of effectively processing such diverse data while maintaining robust feature extraction and learning capabilities. In this paper, we introduce UniFault, a foundation model for fault diagnosis that systematically addresses these issues. Specifically, the model incorporates a comprehensive data harmonization pipeline featuring two key innovations. First, a unification scheme transforms multivariate inputs into standardized univariate sequences. Second, a novel cross-domain temporal fusion strategy mitigates distribution shifts and enriches sample diversity and count, improving the model generalization across varying conditions. UniFault is pretrained on over 6.9 million samples spanning diverse FD datasets, enabling superior few-shot performance. Extensive experiments on real-world FD datasets demonstrate that UniFault achieves state-of-the-art performance, setting a new benchmark for fault diagnosis models and paving the way for more scalable and robust predictive maintenance solutions.

Early detection of patients at risk for MET (Medical Emergency Team) activations is crucial for timely intervention. Threshold-based reference model used in general wards has low sensitivity in detecting clinical deterioration. Its fixed thresholds make early trend prediction difficult, whereas a predicted probability approach provides a more dynamic assessment, where a higher probability indicates a higher risk of deterioration. We describe a machine learning model that enables the identification of patients at risk of clinical deterioration or death 22 h prior to the event, which is 10 h earlier than the current reference (baseline). A retrospective single-center Asian cohort study was conducted from 2019 to 2020 of inpatients in Singapore. A total of 2755 patients with 6496 cases/admissions were included. An XGBoost (XGB) model was trained and tested to compare the performance of the machine learning model to the reference model using sensitivity, specificity, and area under the curve for receiver operating characteristic (AUC-ROC). The XGB model with all vital signs shows highest ROC-AUC (91.7%) and good balance between sensitivity (88.7%) and specificity (88%). The reference model shows highest specificity (98.5%), but very low sensitivity (74.2%) and moderate ROC-AUC (86.4%). The key advantages of machine learning models are they can predict clinical deterioration with higher accuracy (more than 12%) and 10 h earlier than the reference model. In addition, machine learning models require less vital sign inputs for prediction where the reference model requires all vital sign signals. This study also finds that respiration rate and oxygen saturation are primary features in early prediction of clinical deterioration. This aligns with clinical knowledge that abnormalities in these vital signs often serve as early indicators of critical conditions such as respiratory failure or sepsis.

Depression represents a systemic impediment to functional recovery, characterized by multifaceted dysregulation of the neuro-cardiovascular axis. However, traditional assessments often lack the integrative framework required to quantify bidirectional interactions between the central (CNS) and peripheral nervous systems. This study constructs a central-autonomic-blood vascular system bidirectional interaction network (CAB-BiIN) to decode the coupling dynamics underlying these systems. Multi-modal physiological signals were acquired from 60 patients with mild-to-moderate depressive disorder and 54 healthy controls. Information theoretic measures, specifically multi-scale transfer entropy (MSTE), were employed to quantify bidirectional information flow. Our findings demonstrated the frequency-dependent driver differentiation within the physiological networks of depressed individuals. CNS coupling analysis revealed that internal CNS dynamics shift toward suppressed low-frequency (\((\delta)\)) homeostatic regulation, synchronized with the hyperactivity of high-frequency (\((\beta/\gamma)\)) cortical anxiety networks. Furthermore, results indicate a systemic attenuation of rhythmic feedback from both the autonomic and vascular systems to cortical \((\delta)\) oscillations. Parallel to this feedback suppression, a pathologically intensified drive was observed, originating from high-frequency cortical activity and targeting arterial stiffness. These high-frequency neural signals are hypothesized to be transmitted via sympathetic pathways, precipitating sustained vasoconstriction and elevated vascular resistance. By elucidating the symbiotic relationship between physiological dysregulation and psychological symptoms, this study offers a novel quantitative perspective on the pathophysiological mechanisms of depression, and provides a systemic digital diagnostic framework, offering a robust, wearable-based tool for the objective and early-stage diagnosis of depression.

Algorithmic fairness is often studied in static or single-agent settings, yet many real-world decision-making systems involve multiple interacting entities whose multi-stage actions jointly influence long-term outcomes. Existing fairness methods applied at isolated decision points frequently fail to mitigate disparities that accumulate over time. Although recent work has modeled fairness as a sequential decision-making problem, it typically assumes centralized agents or simplified dynamics, limiting its applicability to complex social systems. We introduce , a suite of designed to simulate realistic, modular, and dynamic systems in which fairness emerges from the interplay of multiple agents. We demonstrate MAFEs in three domains—loan processing, healthcare, and higher education—supporting heterogeneous agents, configurable interventions, and fairness metrics. The environments are open-source and compatible with standard multi-agent reinforcement learning (MARL) libraries, enabling reproducible evaluation of fairness-aware policies. Through extensive experiments on cooperative use cases, we demonstrate how MAFE facilitates the design of equitable multi-agent algorithms and reveals critical trade-offs between fairness, performance, and coordination. MAFE provides a foundation for systematic progress in dynamic, multi-agent fairness research.

Deep neural networks deployed in safety-critical applications must provide not only high accuracy but also reliable uncertainty estimates. Deep ensembles are a strong baseline for both, but their computational and storage costs grow linearly with the ensemble size, which limits deployment in resource-constrained settings. Ensemble distillation compresses an ensemble into a single student model; however, standard approaches mainly mimic the ensemble distribution and tend to inherit its residual overconfidence, without explicitly separating calibrated from miscalibrated behavior. We propose Geometric Contrastive Ensemble Distillation (GCED), an ensemble distillation framework that induces a geometric calibration margin in the probability simplex. We begin by analyzing an alignment-repulsion loss and show that it is exactly equivalent to a KL projection onto a tempered geometric ratio constructed from a positive and a negative distribution. Under mild assumptions, this yields a first-order improvement in calibration over standard ensemble distillation. Guided by this analysis, we instantiate GCED with a deep ensemble as the positive teacher, an error-tilted negative distribution that forms an overconfident envelope in miscalibrated regions, and semantic-and feature-level contrastive distillation with a bounded KL-based critic and sample-wise gating. Experiments on image classification and object detection benchmarks show that GCED yields compact students that retain the predictive strength of deep ensembles while consistently improving uncertainty calibration over existing ensemble distillation methods.

Time series domain adaptation aims to transfer the complex temporal dependence from the labeled source domain to the unlabeled target domain. Recent advances leverage the stable causal mechanism over observed variables to model the domain-invariant temporal dependence. However, modeling precise causal structures in high-dimensional data, such as videos, remains challenging. Additionally, direct causal edges may not exist among observed variables (e.g., pixels). These limitations hinder the applicability of existing approaches to real-world scenarios. To address these challenges, we find that the high-dimension time series data are generated from the low-dimension latent variables, which motivates us to model the causal mechanisms of the temporal latent process. Based on this intuition, we propose a latent causal mechanism identification framework that guarantees the uniqueness of the reconstructed latent causal structures. Specifically, we first identify latent variables by utilizing sufficient changes in historical information. Moreover, by enforcing the sparsity of the relationships of latent variables, we can achieve identifiable latent causal structures. Built on the theoretical results, we develop the Latent Causality Alignment (LCA) model that leverages variational inference, which incorporates an intra-domain latent sparsity constraint for latent structure reconstruction and an inter-domain latent sparsity constraint for domain-invariant structure reconstruction. Experiment results on eight benchmarks show a general improvement in the domain-adaptive time series classification and forecasting tasks, highlighting the effectiveness of our method in real-world scenarios.

Recent advancements in machine learning-based signal analysis, coupled with open data initiatives, have fuelled efforts in automatic sleep stage classification. Despite the proliferation of classification models, few have prioritised reducing model complexity, which is a crucial factor for practical applications. In this work, we introduce Multi-Scale and Attention Convolutional Neural Network (MSA-CNN), a lightweight architecture featuring as few as similar to 10,000 parameters. MSA-CNN leverages a novel multi-scale module employing complementary pooling to eliminate redundant filter parameters and dense convolutions. Model complexity is further reduced by separating temporal and spatial feature extraction and using cost-effective global spatial convolutions. This separation of tasks not only reduces model complexity but also mirrors the approach used by human experts in sleep stage scoring. We evaluated both small and large configurations of MSA-CNN against nine state-of-the-art baseline models across three public datasets, treating univariate and multivariate models separately. Our evaluation, based on repeated cross-validation and re-evaluation of all baseline models, demonstrated that the large MSA-CNN outperformed all baseline models on all three datasets in terms of accuracy and Cohen's kappa, despite its significantly reduced parameter count. Lastly, we explored various model variants and conducted an in-depth analysis of the key modules and techniques, providing deeper insights into the underlying mechanisms. The code for our models, baselines, and evaluation procedures is available at https://github.com/sgoerttler/MSA-CNN.

Vector quantization (VQ) with autoregressive (AR) token modeling is a widely adopted and highly competitive paradigm for time-series generation. However, such models are fundamentally limited by exposure bias: during inference, errors can accumulate across sequential predictions, leading to pronounced quality degradation in long-horizon generation. To address this, we propose SDFlow (Similarity-Driven Flow Matching), a non-autoregressive framework that operates entirely in the frozen VQ latent space and enables parallel sequence generation via flow matching. We tackle three key challenges in making this transition: (1) eliminating exposure bias by replacing step-wise token prediction with a global transport map; (2) mitigating the high-dimensionality of VQ token spaces via a low-rank manifold decomposition with a learned anchor prior over the latent manifold; and (3) incorporating discrete supervision into continuous transport dynamics by introducing a categorical posterior over codebook indices within a variational flow-matching formulation. Extensive experiments show that SDFlow achieves state-of-the-art performance, improving Discriminative Score and substantially reducing Context-FID, particularly for challenging long-sequence generation. Moreover, SDFlow provides significant inference speedups over autoregressive baselines, offering both high fidelity and computational efficiency. Code is available at https://anonymous.4open.science/r/SDFlow-D6F3/

Accurate prediction of the Remaining Useful Life (RUL) in machinery can significantly diminish maintenance costs, enhance equipment up-time, and mitigate adverse outcomes. Data-driven RUL prediction techniques have demonstrated commendable performance. However, their efficacy often relies on the assumption that training and testing data are drawn from the same distribution or domain, which does not hold in real industrial settings. To mitigate this domain discrepancy issue, prior adversarial domain adaptation methods focused on deriving domain-invariant features. Nevertheless, they overlook target-specific information and inconsistency characteristics pertinent to the degradation stages, resulting in suboptimal performance. To tackle these issues, we propose a novel domain adaptation approach for cross-domain RUL prediction named TACDA. Specifically, we propose a target domain reconstruction strategy within the adversarial adaptation process, thereby retaining target-specific information while learning domain-invariant features. Furthermore, we develop a novel clustering and pairing strategy for consistent alignment between similar degradation stages. Through extensive experiments, our results demonstrate the remarkable performance of our proposed TACDA method, surpassing state-of-the-art approaches with regard to two different evaluation metrics. Our code is available at https://github.com/keyplay/TACDA.

Multivariate Time-Series (MTS) data is crucial across diverse domains. With its sequential and multi-source (e.g., sensors) properties, MTS data inherently exhibits Spatial-Temporal (ST) dependencies, involving temporal correlations over time and spatial correlations between sensors. While Graph Neural Networks (GNNs) have been widely adopted to exploit these dependencies, existing methods typically capture spatial and temporal dependencies separately, overlooking correlations between Different sEnsors at Different Times (DEDT). Ignoring such correlations limits the comprehensive modeling of ST dependencies and hinders effective MTS representation learning. To address this, we propose a Fully-Connected Spatial-Temporal Graph Neural Network (FC-STGNN), which introduces two key components: FC graph construction and FC graph convolution. In graph construction, we design an FC graph that links all sensors at all times, with edge weights refined according to temporal distances. This design enables comprehensive modeling of ST dependencies by explicitly capturing correlations between DEDT. To exploit this graph, we introduce FC graph convolution with moving-pooling GNN layers, effectively capturing the ST dependencies for MTS representation learning. However, FC-STGNN relies on fixed-size patching, which may limit its ability to capture optimal local patterns for FC graph construction. To overcome this, we extend the framework to GAP-STGNN by incorporating Gaussian Adaptive Patching (GAP), dynamically learning patches with adaptive receptive fields to better capture local patterns. GAP further integrates an adaptive patch selection module that identifies informative patches while softly attenuating less relevant ones to maintain temporal continuity. Extensive experiments on multiple MTS datasets demonstrate that FC-STGNN and GAP-STGNN effectively capture comprehensive ST dependencies with improved FC graphs, achieving superior performance compared to SOTA methods. The code is available at https://github.com/Frank-Wang-oss/GAP-STGNN.

Full parameter fine-tuning of large language models (LLMs) imposes prohibitive computational burdens for edge deployment. Parameter-Efficient Fine-Tuning (PEFT) has emerged as a critical paradigm that optimizes only task-specific parameter subsets rather than the full weight matrix. Among widely used PEFT methods, Low+-Rank Adaptation (LoRA) and its variants have gained considerable popularity as they avoid additional inference costs. However, deterministic LoRA methods remain susceptible to data noise and catastrophic forgetting, creating a persistent accuracy gap with full fine-tuning. To address these limitations, we propose MeLoRA, a probabilistic framework that reformulates the fine-tuning objective using Gaussian variational inference. Building upon established Bayesian principles, our key innovation lies in applying this probabilistic formulation to the low-rank adapter context. MeLoRA treats adapter parameters as random variables governed by learnable Gaussian distributions, thereby shifting optimization from point estimates to the distribution’s mean and covariance. This approach naturally captures parameter uncertainty and acts as an adaptive regularizer, enhancing robustness. We further employ a memory-efficient, low-rank parameterization of the covariance matrix to maintain practical efficiency. Extensive experiments with multiple pre-trained models across Natural Language Understanding (NLU), Natural Language Generation (NLG), and commonsense reasoning tasks validate the effectiveness of MeLoRA. Results show MeLoRA achieves significant improvements over baseline methods, e.g., fine-tuning LLaMA2-7B on the commonsense reasoning benchmark represents a 0.51% relative improvement, while fine-tuning GPT-2 Medium on E2E increases ROUGE-L by 0.6%.

Remaining Useful Life (RUL) prediction for turbofan engines plays a vital role in predictive maintenance, ensuring operational safety and efficiency in aviation. Although data-driven approaches using machine learning and deep learning have shown potential, they face challenges such as limited data and distribution shifts caused by varying operating conditions. Domain Adaptation (DA) has emerged as a promising solution, enabling knowledge transfer from source domains with abundant data to target domains with scarce data while mitigating distributional shifts. Given the unique properties of turbofan engines, such as complex operating conditions, high-dimensional sensor data, and slower-changing signals, it is essential to conduct a focused review of DA techniques specifically tailored to turbofan engines. To address this need, this paper provides a comprehensive review of DA solutions for turbofan engine RUL prediction, analyzing key methodologies, challenges, and recent advancements. A novel taxonomy tailored to turbofan engines is introduced, organizing approaches into methodology-based (how DA is applied), alignment-based (where distributional shifts occur due to operational variations), and problem-based (why certain adaptations are needed to address specific challenges). This taxonomy offers a multidimensional view that goes beyond traditional classifications by accounting for the distinctive characteristics of turbofan engine data and the standard process of applying DA techniques to this area. Additionally, we evaluate selected DA techniques on turbofan engine datasets, providing practical insights for practitioners and identifying key challenges. Future research directions are identified to guide the development of more effective DA techniques, advancing the state of RUL prediction for turbofan engines.

A growing number of techniques leverage the spatial structures that underlie many real-world datasets. Despite these advances, the complementary task of estimating spatial structures and understanding their role within these techniques has often been overlooked. In neurophysiological data analysis specifically, numerous methods exist to estimate brain connectivity, but most are not explicitly model-based, dynamic, multivariate, or directed. To address these limitations, we previously introduced noise-driven heat modelling on graphs for neurophysiological connectivity estimation. In this study, we extend this framework by relaxing earlier noise assumptions and adding regularisation to improve robustness. We also develop a simulation procedure to characterise and evaluate our technique in a controlled setting. Finally, we demonstrate that the technique is able to capture meaningful spatial structure across two experiments, each using two real-world datasets. The explicit model formulation of our connectivity estimator has the potential to improve the interpretability of graph-based techniques across a wide range of applications. The code implementing our method is available at https://github.com/sgoerttler/Heat_Connectivity.

We present the first theoretical convergence analysis of machine learning training under fully homomorphic encryption (FHE), combined with a differentially private (DP) training algorithm tailored to encrypted computation. Our approach improves computational efficiency over standard differentially private gradient descent (DP-GD) while achieving comparable utility. In particular, we prove convergence of approximate gradient descent using polynomial approximations of activation and loss functions, which are required for FHE compatibility. To preserve privacy in downstream tasks, we integrate differential privacy without relying on costly per-sample gradient clipping, enabling scalable encrypted learning. We also provide data-independent hyperparameter selection and theoretically grounded strategies for polynomial approximation which can be of independent interest. Together, these contributions advance the feasibility of efficient, private, and secure machine learning on sensitive data.

Foundation models (FMs) have achieved remarkable success across diverse machine learning domains through large-scale pretraining on large, diverse datasets. However, pretraining on such datasets introduces significant challenges due to substantial mismatches in data distributions, a problem particularly pronounced with time series (TS) data. In this article, we tackle this issue by proposing a domain-aware adaptive normalization strategy within the transformer architecture. Specifically, we replace the traditional LayerNorm with a prototype-guided dynamic normalization mechanism (ProtoNorm), where learned prototypes encapsulate distinct data distributions, and sample-to-prototype affinity determines the appropriate normalization layer. This mechanism effectively captures the heterogeneity of TS characteristics, aligning pretrained representations with downstream tasks. Through comprehensive empirical evaluation, we demonstrate that our method significantly outperforms conventional pretraining techniques across diverse downstream tasks, while effectively mitigating the adverse effects of distribution shifts during pretraining. Incorporating ProtoNorm is as simple as replacing a single line of code. Extensive experiments on diverse real-world TS benchmarks validate the robustness and generalizability of our approach, advancing the development of more versatile TS FMs.

3D content acquisition and creation are expanding rapidly in the new era of machine learning and AI. 3D Gaussian Splatting (3DGS) has become a promising high-fidelity and real-time representation for 3D content. Similar to the initial wave of digital audio-visual content at the turn of the millennium, the demand for intellectual property protection is also increasing, since explicit and editable 3D parameterization makes unauthorized use and dissemination easier. In this position paper, we argue that effective progress in watermarking 3D assets requires articulated security objectives and realistic threat models, incorporating the lessons learned from digital audio-visual asset protection over the past decades. To address this gap in security specification and evaluation, we advocate a scenario-driven formulation, in which adversarial capabilities are formalized through a security model. Based on this formulation, we construct a reference framework that organizes existing methods and clarifies how specific design choices map to corresponding adversarial assumptions. Within this framework, we also examine a legacy spread-spectrum embedding scheme, characterizing its advantages and limitations and highlighting the important trade-offs it entails. Overall, this work aims to foster effective intellectual property protection for 3D assets.

The widespread deployment of high-fidelity generative models has intensified the need for reliable mechanisms for provenance and content authentication. In-processing watermarking, embedding a signature into the generative model's synthesis procedure, has been advocated as a solution and is often reported to be robust to standard post-processing (such as geometric transforms and filtering). Yet robustness to semantic manipulations that alter high-level scene content while maintaining reasonable visual quality is not well studied or understood. We introduce a simple, multi-stage framework for systematically stress-testing in-processing generative watermarks under semantic drift. The framework utilizes off-the-shelf models for object detection, mask generation, and semantically guided inpainting or regeneration to produce controlled, meaning-altering edits with minimal perceptual degradation. Based on extensive experiments on representative schemes, we find that robustness varies significantly with the degree of semantic entanglement: methods by which watermarks remain detectable under a broad suite of conventional perturbations can fail under semantic edits, with watermark detectability in many cases dropping to near zero while image quality remains high. Overall, our results reveal a critical gap in current watermarking evaluations and suggest that watermark designs and benchmarking must explicitly account for robustness against semantic manipulation.

Brain-Computer Interfaces (BCI) have shown promise in facilitating upper limb rehabilitation following stroke. However, restoring fine hand functions, such as grasping, remains a significant challenge. To address this, we focus on decoding hand grasp attempts from electroencephalography (EEG) to enable BCI-driven hand rehabilitation. In this work, we propose several novel methods. First, inspired by the Small-World Brain Network Theory, we introduce a Small-world Hierarchical Interconnected Graph Neural Network (SHINE). SHINE captures transient power dynamics using multiscale convolution, overlapping windows, and learnable variance. It also simulates the characteristic architecture of the brain, where strong local connections coexist with weaker long-range links. This design advances existing Graph Neural Network (GNN) approaches, which typically model functional connectivity using a single, distance-agnostic metric, treating all brain regions uniformly. Second, we propose a Progressive Decay Graph (PDG) mechanism that progressively weakens long-range connections according to distance and training epoch, allowing the model's connectivity structure to evolve alongside the learning process. We evaluated SHINE on two EEG datasets comprising 50 healthy subjects and 19 post-stroke patients performing attempted hand opening and closing, which are the two complementary phases of a grasp. SHINE achieved superior performance over state-of-the-art methods, with improvements of 2.32% (healthy open vs. rest), 1.98% (healthy close vs. rest), 3.97% (stroke open vs. rest), and 2.66% (stroke close vs. rest) ($p< 0.01$ for all tasks), respectively.