Computer Science Review最新文献

This paper provides a comprehensive survey of anomaly detection for the Internet of Things (IoT). Anomaly detection poses numerous challenges in IoT, with broad applications, including intrusion detection, fraud monitoring, cybersecurity, industrial automation, etc. Intensive attention has been received by network security analytics and researchers, particularly on anomaly detection in the network, deliberately crucial in network security. It is of critical importance to detect network anomalies timely. Due to various issues and resource-constrained features, conventional anomaly detection strategies cannot be implemented in the IoT. Hence, this paper attempts to highlight various recent techniques to detect anomalies in IoT and its applications. We also present anomalies at multiple layers of the IoT architecture. In addition, we discuss multiple computing platforms and highlight various challenges of anomaly detection. Finally, the potential future directions of the methods are suggested, leading to various open research issues to be analyzed afterward. With this survey, we hope that readers can get a better understanding of anomaly detection, as well as research trends in this domain.

Fault tolerance is becoming increasingly important for upcoming exascale systems, supporting distributed data processing, due to the expected decrease in the Mean Time Between Failures (MTBF). To ensure the availability, reliability, dependability, and performance of the system, addressing the fault tolerance challenge is crucial. It aims to keep the distributed system running at a reduced capacity while avoiding complete data loss, even in the presence of faults, with minimal impact on system performance. This comprehensive survey aims to provide a detailed understanding of the importance of fault tolerance in distributed systems, including a classification of faults, errors, failures, and fault-tolerant techniques (reactive, proactive, and predictive). We collected a corpus of 490 papers published from 2014 to 2023 by searching in Scopus, IEEE Xplore, Springer, and ACM digital library databases. After a systematic review, 17 reactive models, 17 proactive models, and 14 predictive models were shortlisted and compared. A taxonomy of ideas behind the proposed models was also created for each of these categories of fault-tolerant solutions. Additionally, it examines how fault tolerance capability is incorporated into popular big data processing tools such as Apache Hadoop, Spark, and Flink. Finally, promising future research directions in this domain are discussed.

Hyperspectral image (HSI) classification is a significant topic of discussion in real-world applications. The prevalence of these applications stems from the precise spectral information offered by each pixelś data in hyperspectral imaging (HS). Classical machine learning (ML) methods face challenges in precise object classification with HSI data complexity. The intrinsic non-linear relationship between spectral information and materials complicates the task. Deep learning (DL) has proven to be a robust feature extractor in computer vision, effectively addressing nonlinear challenges. This validation drives its integration into HSI classification, which proves to be highly effective. This review compares DL approaches to HSI classification, highlighting its superiority over classical ML algorithms. Subsequently, a framework is constructed to analyze current advances in DL-based HSI classification, categorizing studies based on a network using only spectral features, spatial features, or both spectral–spatial features. Moreover, we have explained a few recent advanced DL models. Additionally, the study acknowledges that DL demands a substantial number of labeled training instances. However, obtaining such a large dataset for the HSI classification framework proves to be time and cost-intensive. So, we also explain the DL methodologies, which work well with the limited training data availability. Consequently, the survey introduces techniques aimed at enhancing the generalization performance of DL procedures, offering guidance for the future.

Despite the advances in machine learning techniques, similarity assessment among multimedia data remains a challenging task of broad interest in computer science. Substantial progress has been achieved in acquiring meaningful data representations, but how to compare them, plays a pivotal role in machine learning and retrieval tasks. Traditional pairwise measures are widely used, yet unsupervised affinity learning approaches have emerged as a valuable solution for enhancing retrieval effectiveness. These methods leverage the dataset manifold to encode contextual information, refining initial similarity/dissimilarity measures through post-processing. In other words, measuring the similarity between data objects within the context of other data objects is often more effective. This survey provides a comprehensive discussion about unsupervised post-processing methods, addressing the historical development and proposing an organization of the area, with a specific emphasis on image retrieval. A systematic review was conducted contributing to a formal understanding of the field. Additionally, an experimental study is presented to evaluate the potential of such methods in improving retrieval results, focusing on recent features extracted from Convolutional Neural Networks (CNNs) and Transformer models, in 8 distinct datasets, and over 329.877 images analyzed. State-of-the-art comparison for Flowers, Corel5k, and ALOI datasets, the Rank Flow Embedding method outperformed all state-of-art approaches, achieving 99.65%, 96.79%, and 97.73%, respectively.

Lately, a lot of attention has been paid to securing the ownership rights of digital images. The expanding usage of the Internet causes several problems, including data piracy and data tampering. Image watermarking is a typical method of protecting an image's copyright. Robust watermarking for digital images is a process of embedding watermarks on the cover image and extracting them correctly under different attacks. The embedded watermark might be either visible or invisible. Deep learning extracts image features using neural networks, which are highly effective in feature extraction. Watermarking techniques that utilize deep learning have gained a lot of interest due to their remarkable ability to extract features. This article offers an overview of digital image watermarking and deep learning. This article will discuss several research articles on digital image watermarking in deep-learning environments.

The advent of cloud computing has made a global impact by providing on-demand services, elasticity, scalability, and flexibility, hence delivering cost-effective resources to end users in pay-as-you-go manner. However, securing cloud services against vulnerabilities, threats, and modern attacks remains a major concern. Application layer attacks are particularly problematic because they can cause significant damage and are often difficult to detect, as malicious traffic can be indistinguishable from normal traffic flows. Moreover, preventing Distributed Denial of Service (DDoS) attacks is challenging due to its high impact on physical computer resources and network bandwidth. This study examines new variations of DDoS attacks within the broader context of cyber threats and utilizes Artificial Intelligence (AI)-based approaches to detect and prevent such modern attacks. The conducted investigation determines that the current detection methods predominantly employ collectively, hybrid, and single Machine Learning (ML)/Deep Learning (DL) techniques. Further, the analysis of diverse DDoS attacks and their related defensive strategies is vital in safeguarding cloud infrastructure against the detrimental consequences of DDoS attacks. This article offers a comprehensive classification of the various types of cloud DDoS attacks, along with an in-depth analysis of the characterization, detection, prevention, and mitigation strategies employed. The article presents, an in-depth analysis of crucial performance measures used to assess different defence systems and their effectiveness in a cloud computing environment. This article aims to encourage cloud security researchers to devise efficient defence strategies against diverse DDoS attacks. The survey identifies and elucidates the research gaps and obstacles, while also providing an overview of potential future research areas.

A graph vertex-subset problem defines which subsets of the vertices of an input graph are feasible solutions. We view a feasible solution as a set of tokens placed on the vertices of the graph. A reconfiguration variant of a vertex-subset problem asks, given two feasible solutions of size $k$, whether it is possible to transform one into the other by a sequence of token slides (along edges of the graph) or token jumps (between arbitrary vertices of the graph) such that each intermediate set remains a feasible solution of size $k$. Many algorithmic questions present themselves in the form of reconfiguration problems: Given the description of an initial system state and the description of a target state, is it possible to transform the system from its initial state into the target one while preserving certain properties of the system in the process? Such questions have received a substantial amount of attention under the so-called combinatorial reconfiguration framework. We consider reconfiguration variants of three fundamental underlying graph vertex-subset problems, namely Independent Set, Dominating Set, and Connected Dominating Set. We survey both older and more recent work on the parameterized complexity of all three problems when parameterized by the number of tokens $k$. The emphasis will be on positive results and the most common techniques for the design of fixed-parameter tractable algorithms.

Virtual humans have become a hot research topic in recent years due to the development of AI technology and computer graphics. In this survey, we provide a comprehensive review of the interaction design of 3D virtual humans. We first categorize the interac- tion design of virtual humans into speech, eye, facial expressions, and posture interactions. Then we describe the combination of different modalities of virtual humans in the multimodal interaction design section. We also summarize the applications of intelli- gent virtual humans in the fields of education, healthcare, and work assistance. The final part of the paper discusses the remaining challenges and opportunities in virtual human interaction design, along with future directions in this field. This paper hopes to help researchers quickly understand the characteristics of various modal interactions in the process of designing intelligent virtual humans and provide design guidance and suggestions.

Mobile Robots (MRs) and their applications are undergoing massive development, requiring a diversity of autonomous or self-directed robots to fulfill numerous objectives and responsibilities. Integrating MRs with the Intelligent Internet of Things (IIoT) not only makes robots innovative, trackable, and powerful but also generates numerous threats and challenges in multiple applications. The IIoT combines intelligent techniques, including artificial intelligence and machine learning, with the Internet of Things (IoT). The location information (localization) of the MRs triggers innumerable domains. To fully accomplish the potential of localization, Mobile Robot Localization (MRL) algorithms need to be integrated with complementary technologies, such as MR classification, indoor localization mapping solutions, three-dimensional localization, etc. Thus, this paper endeavors to comprehensively review different methodologies and technologies for MRL, emphasizing intelligent architecture, indoor and outdoor methodologies, concepts, and security-related issues. Additionally, we highlight the diverse MRL applications where information about localization is challenging and present the various computing platforms. Finally, discussions on several challenges regarding navigation path planning, localization, obstacle avoidance, security, localization problem categories, etc., and potential future perspectives on MRL techniques and applications are highlighted.

Stock market is one of the attractive domains for researchers as well as academicians. It represents highly complex non-linear fluctuating market behaviours where traders, investors, and organizers look forward to reliable future predictions of the market indices. Such prediction problems can be computationally addressed using various machine learning, deep learning, sentiment analysis, as well as mining approaches. However, the internal parameters configuration can play an important role in the prediction performance; also, feature selection is a crucial task. Therefore, to optimize such approaches, the evolutionary computation-based algorithms can be integrated in several ways. In this article, we systematically conduct a focused survey on genetic algorithm (GA) and its applications for stock market prediction; GAs are known for their parallel search mechanism to solve complex real-world problems; various genetic perspectives are also integrated with machine learning and deep learning methods to address financial forecasting. Thus, we aim to analyse the potential extensibility and adaptability of GAs for stock market prediction. We review stock price and stock trend prediction, as well as portfolio optimization, approaches over the recent years (2013–2022) to signify the state-of-the-art of GA-based optimization in financial markets. We broaden our discussion by briefly reviewing other genetic perspectives and their applications for stock market forecasting. We balance our survey with the consideration of competitiveness and complementation of GAs, followed by highlighting the challenges and potential future research directions of applying GAs for stock market prediction.