Deep learning-based fault localization has recently been widely studied. Previous studies have demonstrated that deep learning-based methods can localize faults more accu-rately than traditional methods, such as spectrum-based and mutation-based approaches. However, many early deep learning-based approaches solely employed code statistic features and achieved subpar performance. Subsequent studies have enhanced performance by incorporating semantic information into the model. Nevertheless, their procedure to extract code semantics is simplistic. Their strategies for integrating semantics with other code features also fail to fully exploit the correlations between features. In this work, we propose FusionFL, a statement-level fault localization method. FusionFL aims to enhance fault localization precision through fine-grained semantic learning and more profound feature fusion. Specifically, FusionFL generates a separate vector representation for each code statement and concatenates SBFL and MBFL scores to construct a suspicious score matrix. Furthermore, FusionFL utilizes an attention mech-anism to capture the relationships among each statement in the matrix and enables the code semantics to learn from these relationships. Finally, FusionFL associates the semantics of each suspicious statement with its corresponding context statements and generates suspicious scores. We evaluate FusionFL on widely used benchmark Defects4j. Experimental results show that FusionFL can effectively fuse multiple features and achieves an average 48.5% improvement in Top-1 over baselines. Meanwhile, we conduct extensive experiments on FusionFL to verify its effectiveness in feature fusion.

With the progress of software technology, vulnerability detection is more important in software security. The mainstream method lacks discriminability and explainability as it uses graph neural network to fuse features simultaneously. Meanwhile only showing the presence of vulnerabilities limits its usefulness. Therefore, we propose BMAAVD, a Vulnerability Detection framework based on Bilateral Masks Aggregate Attention mechanism, including a staged fusion based on feature level with adaptive structures. It treats detection as a multi-classification task and outputs classification information to improve usefulness. BMAAVD performs well on both synthetic and real datasets, achieving an average increase of 7.61 % and 14.82% in F1 score, and the group bagging training strategy we promoted improves the model's performance in data imbalance.

Given an undirected graph, the Maximum k-plex Problem (MKP) is to find a largest induced subgraph in which each vertex has at most k−1 non-adjacent vertices. The problem arises in social network analysis and has found applications in many important areas employing graph-based data mining. Existing exact algorithms usually implement a branch-and-bound approach that requires a tight upper bound to reduce the search space. In this paper, we propose a new upper bound for MKP, which is a partitioning of the candidate vertex set with respect to the constructing solution. We implement a new branch-and-bound algorithm that employs the upper bound to reduce the number of branches. Experimental results show that the upper bound is very effective in reducing the search space. The new algorithm outperforms the state-of-the-art algorithms significantly on real-world massive graphs, DIMACS graphs and random graphs.