Microplastic pollution in global aquatic ecosystems poses an imminent threat to both ecological integrity and human wellbeing. These minuscule particles (<5 mm), derived from anthropogenic activities, accumulate organic pollutants and heavy metals, permeating the food chain and triggering reproductive abnormalities and endocrine disruption in organisms. Particles with diameters smaller than 100 μm are particularly insidious, owing to their diminutive size, which facilitates greater bioaccumulation while rendering them significantly more challenging to collect and detect. Drawing inspiration from the highly efficient filtration mechanism of sabellid worms, this study proposes the design of an aquatic microplastic adsorption robot that mimics the feather-like radiolar crown structure of these organisms. The robot incorporates a flexible polymeric vibrating membrane system, a solid monolithic magnetic porous polymer material (PDVB-Fe₃O₄), and underwater adsorption suction cups to achieve efficient capture of minute microplastics (diameters less than 100 μm). Post-collection, the adsorption module enables rapid desorption, thereby facilitating facile onshore analysis and detection. The authors adopted computational fluid dynamics (CFD) methods to develop a fluid-solid coupling model, simulating five water environments with varying flow velocities: 0.05 m/s, 0.07 m/s, 0.09 m/s, 0.2 m/s, and 0.5 m/s. The results validated the robotic system's in-water performance, revealing low energy consumption and favorable stability (data). This design offers a scalable technical solution for achieving the Sustainable Development Goal 14 (SDG14) target.

Show more

Artificial intelligence regulations typically require that sensitive attributes (such as gender and race) be excluded from algorithmic decision-making to prevent discrimination, a principle commonly referred to as ''fairness through unawareness.'' However, even when sensitive attributes are removed, algorithmic models may still infer such information through proxy variables that are pervasive in data, often via complex nonlinear relationships, thereby perpetuating or even amplifying systemic bias. To address the problem of indirect discrimination under fairness through unawareness, this paper proposes an end-to-end framework that integrates discrimination auditing, diagnosis, and mitigation. First, by incorporating an advanced Transformer-Based Counterfactual Explainer (TABCF), our framework constructs a more reliable bias auditing system capable of accurately uncovering discriminatory behaviors in models. Second, once bias is detected, we introduce an innovative two-stage NOCCO–Shapley diagnostic method that identifies the key proxy variables responsible for discrimination and reveals how the model actually exploits these variables in practice. Finally, to mitigate the identified bias, we implement an adjustable λ-PCF post-processing strategy that enables a quantifiable trade-off between predictive utility and counterfactual fairness without retraining the model. Notably, we find that when the trade-off parameter λ is set to the prior probability distribution of the sensitive attribute in the dataset, the model achieves an optimal balance between fairness and utility. Extensive experiments on four widely used real-world datasets demonstrate that our end-to-end framework not only outperforms existing methods in auditing and diagnosis, but also provides a practical and effective technical pathway for deploying more responsible and fair AI systems in real-world applications.

Show more

In this project, we develop a convolutional neural network (CNN) to classify handwritten digits from the MNIST dataset, a widely used benchmark in computer vision. Unlike traditional image-processing pipelines that rely on engineered features, CNNs automatically learn hierarchical representations directly from raw pixel data. Our model consists of two convolutional layers, max pooling, dropout for regularization, and two fully connected layers. Trained for five epochs using the Adadelta optimizer with learning rate decay, the network achieves a test accuracy of 98.92%. These results demonstrate that even a relatively small CNN can achieve strong performance on MNIST with minimal tuning.

Show more

Accurate valve nozzle detection is an important component of industrial visual inspection systems; however, structural complexity, scale variation, illumination fluctuation, and partial occlusion remain challenging factors that affect detection stability. This study presents YOLOv12-Enhanced, a refined single-stage detection framework developed for industrial valve nozzle scenarios. The proposed approach incorporates three architectural modifications: a RepViT backbone to enhance hierarchical feature representation through structural re-parameterization and global–local modeling, an SPPF module combined with C2PSA attention to strengthen multi-scale contextual feature extraction, and a Global Edge Information Fusion (GEIF) module to integrate shallow edge information with deep semantic features for improved boundary alignment. Experimental evaluation on the Pascal VOC dataset shows that the proposed model achieves 71.0% mAP50 and 54.4% mAP50–95 under identical training conditions, exceeding the baseline YOLOv12n. Ablation experiments further demonstrate that each module contributes incremental performance gains. Evaluation on a self-constructed valve nozzle dataset consisting of 500 real industrial images indicates stable detection behavior under varying illumination and partial occlusion conditions. The experimental findings suggest that the proposed structural refinements provide a balanced enhancement in feature representation and localization precision while maintaining comparable computational complexity.

Show more