Feature Envy is a code smell due to the abnormal calling relationships between methods and classes, which adversely affects software scalability and maintainability.Existing methods mainly use various technologies to model abnormal relationships to detect feature envy.However, these methods only rely on local features such as entity names, which is not robust enough.Moreover, the mining depth of correlation features between entities involved in feature envy is limited.In this paper, we propose a correlation feature mining model based on dual attention to detect feature envy.Firstly, we propose a multi-view-based entity representation strategy, which enhanced the robustness of the model while improving the suitability of the correlation feature and model.Secondly, we add attention mechanism to the channel dimension and spatial dimension of CNN to control the flow of information and capture the correlation features between entities more accurately.Finally, the evaluation results on projects both with and without feature envy injected show that our proposed approach outperforms the state-of-the-art methods.

God class refers to a class that undertakes too many responsibilities for tasks that should more appropriately be handled by multiple classes. The existence of god classes seriously affects the maintainability and understandability of software. To eliminate god class, we first need to identify them. Researchers have proposed traditional methods using code metrics and deep learning methods using code metrics and text information to detect god classes. However, the relationship existing in metrics and text information is often ignored; moreover, deep learning methods require a large number of reliable datasets, while authentic god class datasets are scarce. To solve the above problems, we propose a novel god class detection method based on multi-aspect interactions and dataset fine-tuning. First, we use proposed model to extract multi-aspect interaction information, including three parts: (i) the interaction information existing in code metrics; (ii) the interaction information existing in texts; (iii) the interaction information existing in texts and code metrics. In this way, we can not only make use of code metrics and text information, but also fully exploit the multi-aspect interaction information. Second, we train with large-scale synthetic datasets to obtain a pre-trained model, then fine-tune the pre-trained model parameters with high-quality authentic datasets. Using the training method of pre-training and fine-tuning, we can solve the problem of low-reliability synthetic datasets and scarce authentic datasets. Finally, evaluation results on open-source applications suggest that the proposed approach improves on the state-of-the-art.

Code smell detection can help developers identify position of code smell in projects and enhance the quality of software system. Usually codes with similar semantic relationships have greater code dependencies, and most code smell detection methods ignore dependencies relationships within the source code. Thus, their detection results may be heavily influenced by inadequate code feature, which can lead to some code smell not being detected. In addition, existing methods cannot explain the correlation between detection results and code information. However, an explainable result can help developers make better judgments on code smell reconstruction. Accordingly, in this paper, we propose a local and global feature based explainable approach to detecting feature envy, one of the most common code smells. For the model to make the most of code information, we design different representation models for global code and local code respectively to extract different feature envy features, and automatically combine these features that are beneficial in terms of detection accuracy. We further design a code semantic dependency (CSD) to make the detection result easy to explain. The evaluation results of seven manual building code smell projects and three real projects show that the proposed approach improves on the state-of-the-art in detecting feature envy and boosting the explainability of results.

As point cloud is a typical and significant type of geometric 3D data, deep learning on the classification and segmentation of point cloud has received widely interests recently. However, the critical problems to process the irregularity of point cloud and feature extraction of shape pattern have not yet been fully explored. In this paper, a geometric deep learning architecture based on our PointDoN module is presented. Inspired by the Difference of Normals (DoN) in traditional point clouds processing, our PointDoN module is a feature aggregation module combining DoN shape pattern descriptor with both 3D coordinates and extra features (such as RGB colors). Our PointDoN-based architecture can be flexibly applied to multiple point cloud processing tasks such as 3D shape classification and scene semantic segmentation. Experiments demonstrate that PointDoN model achieves state-of-the-art results on multiple types of challenging benchmark datasets.

Artificial intelligence (AI) is enabling the automated analysis of large amounts of image/video data, boosting the speed of multimedia data processing remarkably. Meanwhile, Image Quality Assessment (IQA) plays an important role in developing automatic analysis methods. To ensure the effectiveness of AI, images in multimedia applications should be considered for visual examination by both human and machine. Therefore, it is significant to understand the differences between human's and AI's perception of semantic distortion. However, little work has been done due to the lack of data from human on the semantic level. In this paper, we first propose a semantic database (SID) based on the surveillance scenarios, by collecting subjective average recognition rates of 3 semantic targets (face, pedestrian, license plate) with 3 types of distortion (JPEG Compression, BPG Compression, Motion Blur). Then, we present a detailed analysis of how human and AI perceive semantic distortion differently. Experimental results show that AI is stronger in tolerance to distortion than human beings on average, while weaker at generalization and stability. It is also implied in the experiments that existing IQA methods are not effective enough at judging the semantic distortion.

There are a lot of middle-class or small-class high-performance computing clusters at universities and research institutes, etc. Large volumes of job logs have been accumulated after many years of operation. In this paper, on the basis of accumulated job logs on a high-performance computing cluster, we examine and analyze the job logs. Then, we study machine learning based performance analysis and prediction methods for parallel jobs. Various machine learning methods such as multivariate linear fitting, artificial neural network are used to build performance prediction models. We compare the errors of each model, and select the optimal prediction model for different users. The experimental results show that we can obtain reasonable prediction accuracy using the selected machine learning algorithms.

Semantic information is important in video encryption. However, existing image quality assessment (IQA) methods, such as the peak signal to noise ratio (PSNR), are still widely applied to measure the encryption security. Generally, these traditional IQA methods aim to evaluate the image quality from the perspective of visual signal rather than semantic information. In this paper, we propose a novel semantic-level full-reference image quality assessment (FR-IQA) method named Semantic Distortion Measurement (SDM) to measure the degree of semantic distortion for video encryption. Then, based on a semantic saliency dataset, we verify that the proposed SDM method outperforms state-of-the-art algorithms. Furthermore, we construct a Region Of Semantic Saliency (ROSS) video encryption system to demonstrate the effectiveness of our proposed SDM method in the practical application.