Due to the superiority of convolutional neural networks, many deep learning methods have been used in image classification. The enormous difference between natural images and remote sensing images makes it difficult to directly utilize or modify existing CNN models for remote sensing scene classification tasks. In this article, a new paradigm is proposed that can automatically design a suitable CNN architecture for scene classification. A more efficient search framework, RS-DARTS, is adopted to find the optimal network architecture. This framework has two phases. In the search phase, some new strategies are presented, making the calculation process smoother, and better distinguishing the optimal and other operations. In addition, we added noise to suppress skip connections in order to close the gap between trained and validation processing and ensure classification accuracy. Moreover, a small part of the neural network is sampled to reduce the redundancy in exploring the network space and speed up the search processing. In the evaluation phase, the optimal cell architecture is stacked to construct the final network. Extensive experiments demonstrated the validity of the search strategy and the impressive classification performance of RS-DARTS on four public benchmark datasets. The proposed method showed more effectiveness than the manually designed CNN model and other methods of neural architecture search. Especially, in terms of search cost, RS-DARTS consumed less time than other NAS methods.

Recent years have witnessed the growing research interest in the Context-Aware Recommender System (CARS). Specifically, CARS for Web service provides opportunities for exploring the important role of temporal and spatial contexts, separately. Although many CARS approaches have been investigated in recent years, they do not fully address the potential of temporal-spatial correlations in order to make personalized recommendation. In this paper, the Context-Aware Services Recommendation based on Temporal-Spatial Effectiveness (named CASR-TSE) method is proposed. We first model the effectiveness of spatial correlations between the user's location and the service's location on user preference expansion before the similarity computation. Second, we present an enhanced temporal decay model incorporating the weighted rating effect to improve the prediction accuracy in similarity computation. Finally, we evaluate the CASR-TSE method on a real-world Web services dataset. Experimental results show that the proposed method significantly outperforms existing approaches, and thus it is much more effective than traditional recommendation techniques for personalized Web service recommendation.

As one of the most prevalent mental disorders, depression is associated with a high rate of self-harm and suicide, particularly among college students. It is urgently needed to discover prospective cases of depression disorder among college students, enabling timely intervention to reduce its impact on their academic performance and daily lives. This study investigates a method for identifying groups that may have early depressive tendencies through their Internet usage on campus networks. This article proposes a heterogeneous graph attention network (H-GAT) model that incorporates an attention mechanism based on ablation experiments in heterogeneous graphs to analyze the patterns and correlations within the surfing behavior data of students. This model makes full use of the interaction relationships between heterogeneous nodes in the graph to capture the affective tendencies reflected in the cyber-activity patterns. The proposed H-GAT model exhibits excellent performance, with nearly 80% accuracy and recall. Our work offers a potential approach to detect depression on college campuses using nonintrusive methods, which could ultimately contribute to early warnings for both individuals experiencing depression and higher education institutions.