Abstract
Background: Post-Intensive Care Syndrome (PICS), comprising physical, cognitive, and psychological impairments, affects 50-75% of Intensive Care Unit (ICU) survivors and leads to long-term deficits. Virtual Reality (VR) has emerged as a tool to reduce ICU-related stress and support recovery, yet evidence remains fragmented and heterogeneous. Objective: To systematically review the safety, feasibility, and effects of immersive VR interventions targeting PICS-related outcomes in ICU and post-ICU populations, and to introduce a standardized taxonomy to classify and compare VR interventions in critical care contexts. Methods: This systematic review followed PRISMA 2020 guidelines and was registered in PROSPERO (CRD420251174623). Seven databases (Cochrane Library, PubMed, ScienceDirect, IEEE Xplore, ACM Digital Library, SpringerLink, and Scopus) were searched from inception to 2 August 2025. Eligible studies included ICU patients receiving immersive VR via head-mounted displays and targeting at least one PICS domain. Two reviewers independently screened studies and extracted data. Methodological quality was assessed using the Mixed Methods Appraisal Tool (MMAT, 2018). Due to substantial heterogeneity, findings were synthesized narratively. Results: Eleven studies were included. The most consistent effects concerned acute psychological outcomes, with 63.6% of studies reporting reduced anxiety or distress. Evidence for physical, cognitive, or long-term outcomes was limited and inconsistent, largely due to small samples, non-randomized designs, and brief intervention dosing. Conclusion: Current evidence supports VR as a feasible adjunct for acute psychological support in ICU settings. However, meaningful rehabilitation effects remain underexplored. The Five-Axis VR-PICS taxonomy clarifies intervention heterogeneity and provides a structured framework to guide rehabilitation-oriented VR research in critical care.

Abstract
Multi-label image classification datasets are often partially labeled where many labels are missing, posing a significant challenge to training accurate deep classifiers. Most existing approaches assume the missing labels as negatives and/or exploit image and category relationships to regularize training. Orthogonally, this paper studies blending samples in such incomplete datasets as new samples, extending the training data magnitude to increase generalization. First, the proposed LogicMix mixes multiple partially labeled samples to produce new samples, where their unknown labels are naturally mixed by OR's logical equivalences, without replacement with constants. Subsequently, a Decouple Partial-Asymmetric Loss is proposed to assign separate label-focusing policies to original and new samples, addressing the learning imbalance from the different positive-negative label imbalances between original and augmented samples. Finally, we propose a complete learning framework called 2WayAug-PL. LogicMix and conventional data augmentation collaborate to extend the diversity of new samples in both the sample-sample relation and human prior knowledge, while pseudo-labeling compensates for the lack of labels to provide more supervision signals. 27 partially labeled dataset scenarios derived from three benchmarking datasets with various learning difficulties are utilized for comprehensive experiments. LogicMix has shown remarkable effectiveness and generality in improving mAP against compared sample-mixing data augmentation methods. In particular, 2WayAug-PL achieves state-of-the-art average mAP of 84.3%, 50.1 %, and 93.8% on MS-COCO, VG-200, and Pascal VOC 2007, respectively. It further pushes the previous best performance achieved by different frameworks by 0.6% (CFT), 0.6% (CFT), and 0.1 % (SR). Moreover, 2WayAug-PL significantly outperforms all compared frameworks, as shown by statistical tests. Code is available at: https://github.com/maxium0526/logic_mix.

Abstract
The video game industry has grown to become one of the largest in the market, surpassing even the film industry over a decade ago (Statista in Video game industry revenue worldwide 2000-2020). However, the development of games designed with visually impaired players in mind is still almost non-existent when compared to the sheer number of games released yearly. NonVisual Pong is our approach to addressing this challenge, providing blind players with a way to engage in competitive fun through gaming. We took the original Pong game from 1972 and fully adapted it to be played using only a controller-no visual display required. Following the development process, we tested our implementation with experts, discovering that, overall, our game was easy to pick up, required no overly complex setup, and successfully delivered the intended experience. Players enjoyed a balanced challenge and immersion, facilitated by audio cues and the controller's vibrations.

Abstract
Multi-label image classification models often inevitably learn on partially labeled datasets, where a considerable proportion of labels are missing. However, the popular PyTorch deep learning ecosystem is less compatible with training on partially labeled datasets, as many built-in functions like loss functions and metrics do not work correctly or raise errors when unknown labels are present. To this end, we present an original and easy-to-install Python package called mlcpl, which expands the PyTorch ecosystem to offer a friendly environment for learning with partially labeled datasets. The package provides a series of multi-label loss functions and metrics that are compatible with unknown labels. Seven recently proposed approaches are also implemented for the convenient use of cutting-edge techniques. In addition, eleven dataset loading functions, followed by three partial label simulation schemes, expedite the development of experiments. Furthermore, these functions are simple to use, have a PyTorch-like interface, and can collaborate well with other PyTorch components. Several examples of experiments with mlcpl are also provided for demonstration. We wish the release of this package could facilitate relevant academic research and real-world applications. The source code is available at https://github.com/ maxium0526/mlcpl.

Abstract
Solar photovoltaic (PV) power generation has become one of the most widely adopted forms of clean energy worldwide. In large-scale PV farm operation and maintenance, unmanned aerial vehicles equipped with thermal infrared (TIR) cameras are increasingly used to enable automated fault detection and classification. However, the long imaging distance and the inherently low resolution of TIR images often lead to fault patterns appearing with low contrast, making subtle discriminative features difficult to extract and posing significant challenges to achieving highly accurate fault identification and classification. To address these challenges, we propose GEPFNet, a network that exploits Group Equivariant Convolutions to explicitly model the geometric structures of faults, incorporates multi-scale processing with unified local-global contextual representations, and adopts a parallel feature fusion strategy to integrate multi-level features and enhance contextual utilization effectively. The design of feature extraction and fusion mechanisms ensures the proposed GEPFNet achieves strong robustness and generalization under complex operational conditions. The effectiveness of GEPFNet was validated on two public datasets with distinct resolutions, class distributions, and feature characteristics: PVF-10 and the Infrared Solar Module (ISM) dataset. Extensive experiments and statistical analyses demonstrate that the proposed GEPFNet achieves state-of-the-art performance on the PVF-10 dataset, obtaining an accuracy of 96.05 %+/- 0.42 for the 2-Class task and 94.64 %+/- 0.35 for the 10-Class task. On the ISM dataset, GEPFNet achieves an improvement of approximately 5 % over the baseline models. Moreover, under highly imbalanced data distributions, the proposed GEPFNet achieves average accuracy improvements of 5.83% and 3.82% on PVF-10 and ISM, respectively, further demonstrating its capability to enhance class-wise performance. With only 9.51 GFLOPs, GEPFNet also exhibits notable computational efficiency, making it well suited for PV fault classification in TIR imagery.

Abstract
Wikipedia is the largest and most globally well-known online encyclopedia, but its collaborative nature leads to a significant disparity in article quality. In this work, we explore real-time and automatic quality assessment within Wikipedia through machine-learning. We first constructed a dataset of 36,000 English articles and 145 features, then compared the performance of multiple classification and regression algorithms and studied how the number of classes and features affects the model's performance. The six-class experiments achieved a classifier accuracy of 64% and a mean absolute error of 0.09 in regression methods, which matches or beats most state-of-the-art approaches. Our model produces similar results on some non-English Wikipedias, but the error is slightly higher on other versions. We have also determined that the features measuring the article's content and revision history bring the largest performance boost.