Data Analysis and Mining: New Techniques and Applications
A special issue of Applied Sciences (ISSN 2076-3417). This special issue belongs to the section "Computing and Artificial Intelligence".
Deadline for manuscript submissions: 20 August 2024 | Viewed by 5779
Special Issue Editor
Interests: data mining; social network analysis; multimodel learning; graph data analysis; time serial analysis
Special Issue Information
Dear Colleagues,
Learning hierarchical representation and finding useful patterns from data by differentiable models in an end-to-end fashion has been amongst of the greatest developments in data mining so far. Despite its application in traditional research fields like computer vision, natural language processing, and recommendation systems, such a data-driven approach shows great potential when it comes to the intersection of AI and science. From protein structure prediction to quantum artificial intelligence, data mining techniques are providing amazing insight into fitting data and have assisted in the discovery of scientific laws in various domains, as well as contributing to a new research paradigm called AI for science.
Even though artificial general intelligence (AGI) is far from reach, mining scientific data still find many intriguing applications. Recent applications include, but are not restricted to, quantum physics, computational chemistry, molecular biology, fluid dynamics, software engineering, and other disciplines. This Special Issue invites the submission of papers with innovative ideas either in data mining algorithms or in applications of a specific research field. To facilitate the application of data mining technology and accelerate the process of its industrial application, papers that present data mining tools in a specific domain are also welcomed.
Dr. Donghai Guan
Guest Editor
Manuscript Submission Information
Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.
Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Applied Sciences is an international peer-reviewed open access semimonthly journal published by MDPI.
Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2400 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.
Keywords
- data mining
- time series analysis
- multimodel learning
- social network analysis
- classification
- clustering
- graph data analysis
Planned Papers
The below list represents only planned manuscripts. Some of these manuscripts have not been received by the Editorial Office yet. Papers submitted to MDPI journals are subject to peer-review.
Title: Software Defect Prediction with Semantic, Context Features and multi-adaptation
Authors: Chuanqi Tao
Affiliation: taochuanqi@nuaa.edu.cn
Abstract: Many software testing methods, such as random testing and other methods, have been extensively used, but these testing methods may result in a lot of waste of resources. Software defect prediction (SDP), which predicts defective code regions, can help developers find errors and make reasonable testing plans. Cross-project defect prediction(CPDP) model is mainly learning through sufficient data and labels of other projects. Then predicting the defective label of another new project with insufficient data and few labels. Although CPDP has great advantages when there is little historical data of the new project, previous methods are mainly designed with handcrafted features and semantic features. However source code contains rich information including semantic and context features and it is important to know code’s context features in order to diagnose defective code, in this paper we combine handcrafted features with semantic and context features from source code and utilize them conducting experiments on both with-project defect prediction(WPDP) and CPDP. And existing CPDP methods based on the deep learning model have not fully considered the differences among projects and the domain multi-adaption method. To solve these problems, the authors propose a model to automatically generate semantic and context features from source code and then utilize joint domain adaption with multi-layer and multi-kernel maximum mean discrepancy (MLMK-MMD) in deep transfer learning for CPDP.
Title: Pcilad:Pre-trained Temporal Spatial Network for UUV Anomaly Trajectory Detection
Authors: Donghai Guan
Affiliation: College of Computer Science & Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China
Abstract: The recognition of abnormal trajectory in unmanned underwater vehicles(UUV) is crucial for navigation safety and efficiency. Existing works mainly rely on machine learning and probability density, which are difficult to learn the spatiotemporal information of trajectory data, resulting in low abnormal recognition rates and poor transferability across different tasks. To address this issue, this paper proposes a multi-dimensional spatiotemporal fusion model named Pcilad which leverages pre-training techniques and Pcilad is designed to learn spatiotemporal information and enhance transferability through pre-training and fine-tuning. In the pre-training stage, a spatiotemporal encoder-decoder architecture was utilized to extract spatiotemporal features of trajectory sequences. To capture the spatiotemporal dependencies of UUV trajectory, the sequence was dynamically masked and randomly embedded through masked spatiotemporal trajectory modeling. In the fine-tuning stage, the pre-trained spatiotemporal encoder weights were loaded into classifiers in downstream tasks for end-to-end fine-tuning. This paper conducted experiments on five datasets, and results showed that Pcilad could significantly improve abnormal recognition rates and outperform existing models in terms of performance.