The integration of machine learning (ML) into OS kernels is severely hampered by the high latency of offloading to discrete GPUs (dGPUs), where data transfers across the PCIe bus can consume over 93% of the total execution time. This paper argues that for many latency-sensitive kernel tasks, the solution is not a more powerful dGPU but a fundamental shift to an I/O-efficient architecture: the integrated GPU (iGPU) found in modern APUs.

The persistent challenges of data sparsity and class imbalance have long limited the development of recommender systems. Fortunately, line graph theory offers a novel perspective to overcome these issues. By transforming the user-item interaction bipartite graph into a line graph, the problems of data sparsity and class imbalance are elegantly reformulated as those of insufficient labeled nodes and imbalanced label distribution in the line graph domain. This reformulation allows us to directly apply mature techniques from node classification and imbalanced graph learning to address these core challenges. Inspired by this insight, we propose a Line Graph Data Augmentation (LGDA) strategy, which features two distinct characteristics. Firstly, it is a plug-and-play module that resolves data sparsity and imbalance without modifying the underlying recommendation framework. Secondly, it employs a targeted augmentation and confidence filtering mechanism to generate high-quality, balanced augmented data. Extensive experiments on four real-world datasets validate that LGDA effectively alleviates data sparsity and class imbalance, leading to significant improvements in both recommendation performance and system robustness.

Link prediction is an important task that has numerous applications, including in recommender systems and social network analysis. Autoencoder is an effective method for solving the link prediction task. However, most existing autoencoder-based methods neither fully utilize the attribute information of the nodes nor fully take into account the potential data distribution in the graph. In this article, we propose a novel method named attribute-augmented adversarial variational graph autoencoder (A${}^{3}$VGAE), which can effectively solve the above two problems. The method first constructs the attribute structure graph based on the attribute information. Then, it inputs the topological structure graph, the attribute structure graph, and the attribute information into the shared encoder to obtain two latent representations. Besides, the topology structure graph, attribute structure graph, and attribute information are reconstructed by the dual decoder. The adversarial mechanism is introduced to ensure that the two latent representations match specific prior distributions. Extensive experiments conducted on four real-world graph datasets demonstrate the superiority of our proposed A${}^{3}$VGAE in link prediction tasks.

Identifying positive influence dominating set (PIDS) with the smallest cardinality can produce positive effect with the minimal cost on a social network. The purpose of this article is to propose new approximation algorithms for the minimum PIDS problem and its variants such as the minimum connected PIDS and the minimum PIDS of multiplex networks, with the aim of finding target sets with smaller cardinality. Through the design of novel submodular potential function, we theoretically prove that new approximation algorithms yield approximation ratios with same order compared with existing algorithms. We further demonstrate the performance of our algorithm by showcasing its efficacy on several real-world and publicly available instances of social networks, thereby providing additional evidence that our proposed algorithm can identify PIDS with smaller cardinality.

Knowledge graphs (KGs) are structured knowledge bases that represent information as a collection of interconnected entities and relations. Link prediction in KGs aims to infer missing or potential links between entities based on triple facts. Among different link prediction methods, knowledge graph embedding (KGE) has gained widespread popularity, with the goal of learning low-dimensional representations for KGs. However, most present KGE methods struggle to capture both local and global neighborhood information efficiently. Additionally, many hybrid methods have limitations in modeling and capturing interactions between triples. In this article, we propose an entity-relation-guided random walk (ERGRW) method for link prediction in KGs. Unlike conventional approaches that solely focus on entity-based walks, ERGRW creatively introduces relations as objects to walk as well. Inspired by distance-based methods, we design novel random walk rules based on the translation principle within triples. Thus, the ERGRW not only captures local and global neighborhood information but also discovers potential semantic relationships and interactions in the KGs. Furthermore, the encoder–decoder framework of ERGRW is able to learn comprehensive representation and improve link prediction performance. Extensive experiments conducted on four standard datasets demonstrate the superiority of ERGRW for link prediction.

Community detection is a fundamental problem in complex network analysis that aims to find closely related groups of nodes. Recently, network embedding techniques have been integrated into community detection in two manners to capture the intricate relationships between nodes. The two-staged manner generates node embedding vectors and obtains communities by running a clustering algorithm on them. The single-staged manner simultaneously obtains node embedding vectors and communities by optimizing a hybrid objective concerning with node–community relationships. The general-purpose network embedding algorithms used in the first manner do not emphasize retaining node–community relationships. The second manner ignores the influence of a node’s location in a community (at the center or boundary) and its attributes on community generation. In this article, we propose a biased-random-walk-based community detection (BRWCD) algorithm to tackle the issues. First, a topology-weighted degree is designed to enhance the random walk at the boundary of and inside a community to extract communities precisely. Second, we design an attribute-to-node influence index and an attribute-weighted degree to distinguish different attributes’ influence on node transition to obtain communities with high internal cohesion. Comprehensive experiments on the real-world and synthetic networks demonstrate that BRWCD achieves nearly 10% higher accuracy at most than the state-of-the-art algorithms.

In today's society, social networking has been integrated into everyone's life. The detection of network community has been a hotspot in recent years, and it has been widely used in fraud prevention, personalized recommendation, risk control and other fields. In this paper, we present an improved community detection algorithm for fusing topology and attribute information (FTAI) based on Non-Negative Matrix Factorization (NMF). Firstly, we use the attribute similarity matrix instead of a binary matrix based on string matching to cope with the sparsity of the attribute matrix. Next, we introduce the transfer matrix and the attribute feedback adjustment method to fuse the topological matrix and the attribute matrix. Then, we deduce the iteration formula of each matrix with a rigorous mathematical method, which shows the reliability of the algorithm. Finally, we evaluate our method with extensive experiments by using the data set from a real academic social network (SCHOLAT). Experimental results show that FTAI is superior to other baseline algorithms. The results also indicate that FTAI is more flexible, robust and suitable for community detection in social networks with complex attribute information.