Search

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

In the ever-evolving landscape of cyber threats, tactics, techniques and procedures (TTPs) play a crucial role in understanding malicious activities, providing a fine-grained perspective on the status of cybersecurity, and comprehensively illustrating cyber attack behaviors. Despite significant research efforts in the field of automated identification and extraction of TTPs, a comprehensive systematic review is currently lacking. This paper presents an in-depth analysis of the progress in this area by employing three principal approaches : traditional natural language processing, machine learning, and large language models. The study categorizes the tasks into information extraction, text classification, and text generation, and presents a summary of the general framework for identification and extraction processes. It offers a clear scope of unstructured text and TTPs, while refining the processing and analysis procedures, as well as innovative directions for each approaches. Moreover, building upon existing research, the paper identifies current challenges and proposes future research directions and development opportunities. This comprehensive survey serves as a valuable literature review to support readers in applying advanced technologies and methods for advancing research in this field. [ABSTRACT FROM AUTHOR]

Domain adaptation, the important branch of transfer learning, aims to solve the problem that the performance of traditional machine learning algorithms drops sharply when the training and test samples obey different data distributions. Transformer is a deep learning framework based on a self-attention mechanism, which has strong global feature extraction ability and modeling ability. In recent years, the combination of Transformer and domain adaptation has also become a research hotspot. Although many relative methods have been published, the review of Transformer application in domain adaptation has not been reported. In order to fill the gap in this field and provide reference for relevant research, this paper summarizes and analyzes some typical domain adaptation methods based on Transformer in recent years. This paper summarizes the concepts related to domain adaptation and the basic structure of the Transformer, sorts out various domain adaptation methods based on Transformer from four applications, i.e. image classification, image semantic segmentation, object detection and medical image analysis and compares the domain adaptation methods in image classification. Finally, the challenges of the current domain adaptation Transformer model are summarized, and the feasible research directions in the future are discussed. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

The key to driverless technology is that the decision-making level makes accurate instructions based on the input information of the perception link. Reinforcement learning and imitation learning are better suited for complex scenarios than traditional rules. However, the imitation learning represented by behavioral cloning has the problem of composite error, and this paper uses the priority empirical playback algorithm to improve the behavioral cloning to improve the fitting ability of the model to the demo dataset. The original DDPG (deep deterministic policy gradient) algorithm has the problem of low exploration efficiency, and the experience pool separation and random network distillation (RND) technology are used to improve the DDPG algorithm and improve the training efficiency of DDPG algorithm. The improved algorithm is used for joint training to reduce the useless exploration in the early stage of DDPG training. Verified by TORC (the open racing car simulator) simulation platform, the experimental results show that the proposed method can explore more stable road maintenance, speed maintenance and obstacle avoidance ability within the same number of training times. [ABSTRACT FROM AUTHOR]

Text classification is a basic task in natural language processing, which has important applications in sentiment analysis, news classification and other fields. Compared with traditional machine learning and deep learning models, prompt learning can construct prompts for text classification in the case of insufficient data. In recent years, the emergence of GPT-3 has promoted the development of cue learning methods, and has made significant progress in the field of text classification. Firstly, this paper briefly combs the process of previous text classification methods and analyzes their existing problems and shortcomings. Secondly, it expounds the development process of cue learning and the method of constructing cue templates, and introduces and summarizes the research and results of cue learning methods for text classification. Finally, the development trend and difficulties to be further studied in the field of text classification are summarized and prospected. [ABSTRACT FROM AUTHOR]

The sample balance is crucial for machine learning. The importance of certain classes may be higher than its number on the imbalanced datasets. This paper studies the imbalanced datasets for speech emotion recognition. Firstly, the imbalanced baseline datasets EMODB and IEMOCAP are augmented with different signal-to-noise ratios, and the datasets EMODBM and IEMOCAPM are constructed. Secondly, six techniques namely SMOTE, RandomOverSampler, SMOTEENN, ADASYN, TomekLinks and SMOTETomek are adopted to resample the baseline datasets, and the augmented datasets are constructed to achieve the category balance. Thirdly, 21-dimensional low-level descriptor features are extracted from the baseline datasets and the augmented datasets. Finally, a novel model MA-CapsNet is proposed to validate the effectiveness of the resampling techniques. The results show that all types of emotion samples are basically balanced after resampling, which makes the learning of the model MA-CapsNet fairer. In addition, the model MA-CapsNet has better robustness on the resampling datasets. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)