IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-05 DOI: 10.1016/j.array.2025.100630

Sayeed Salih , Omayma Husain , Refan Mohamed Almohamedh , Hayfaa tajelsier , Aisha Hassan Abdalla Hashim , Hashim Elshafie , Abdelwahed Motwakel

Generative Artificial Intelligence (GAI) is revolutionizing digital marketing by auto-content creation, personalized customer experience, and data-driven decisions. This study conducts a systematic literature review and case study analysis to explore GAI applications, benefits, and challenges in modern digital marketing. Drawing on an extensive analysis of academic journals and industry publications, the current research examines leading GAI software such as ChatGPT, DALL-E, MidJourney, Jasper.ai, and Synthesia based on how they aid in content creation, visual design, and video production. The research also provides real-world case studies in multiple industries, such as retail and fashion, food and beverages, and travel and tourism. The case findings illustrated how GAI augments marketing automation, facilitates customer engagement, and amplifies brand engagement, resulting in greater customer satisfaction, higher conversion rates, and better campaign performance. Although it has several benefits, the adoption of GAI is hampered by several critical barriers, such as data privacy, ethical risks, worker resistance, quality control issues, and infrastructure constraints. This research pinpoints these essential challenges and offers practical solutions. It provides actionable insights for businesses seeking to leverage GAI for competitive advantage in the evolving digital landscape by bridging the gap between theory and practice. The findings contribute to the growing discourse on AI-driven marketing strategies and lay the foundation for future research on GAI's long-term impact on consumer engagement and brand loyalty.

生成式人工智能（GAI）通过自动内容创建、个性化客户体验和数据驱动决策，正在彻底改变数字营销。本研究通过系统的文献回顾和案例分析，探讨GAI在现代数字营销中的应用、效益和挑战。通过对学术期刊和行业出版物的广泛分析，目前的研究考察了ChatGPT、DALL-E、MidJourney、Jasper等领先的GAI软件。ai和synia基于它们如何帮助内容创作、视觉设计和视频制作。该研究还提供了多个行业的真实案例研究，如零售和时尚、食品和饮料、旅游和旅游业。案例结果说明了GAI如何增强营销自动化，促进客户参与，并扩大品牌参与，从而提高客户满意度，提高转化率和更好的活动效果。尽管GAI有几个好处，但它的采用受到几个关键障碍的阻碍，例如数据隐私、道德风险、工人抵制、质量控制问题和基础设施限制。这项研究指出了这些基本挑战，并提供了切实可行的解决方案。它通过弥合理论与实践之间的差距，为寻求利用GAI在不断发展的数字环境中获得竞争优势的企业提供了可操作的见解。这些发现有助于对人工智能驱动的营销策略进行越来越多的讨论，并为未来研究人工智能对消费者参与和品牌忠诚度的长期影响奠定基础。

{"title":"From ideation to execution: Unleashing the power of generative AI in modern digital marketing and customer engagement- A systematic literature review and case study","authors":"Sayeed Salih , Omayma Husain , Refan Mohamed Almohamedh , Hayfaa tajelsier , Aisha Hassan Abdalla Hashim , Hashim Elshafie , Abdelwahed Motwakel","doi":"10.1016/j.array.2025.100630","DOIUrl":"10.1016/j.array.2025.100630","url":null,"abstract":"<div><div>Generative Artificial Intelligence (GAI) is revolutionizing digital marketing by auto-content creation, personalized customer experience, and data-driven decisions. This study conducts a systematic literature review and case study analysis to explore GAI applications, benefits, and challenges in modern digital marketing. Drawing on an extensive analysis of academic journals and industry publications, the current research examines leading GAI software such as ChatGPT, DALL-E, MidJourney, Jasper.ai, and Synthesia based on how they aid in content creation, visual design, and video production. The research also provides real-world case studies in multiple industries, such as retail and fashion, food and beverages, and travel and tourism. The case findings illustrated how GAI augments marketing automation, facilitates customer engagement, and amplifies brand engagement, resulting in greater customer satisfaction, higher conversion rates, and better campaign performance. Although it has several benefits, the adoption of GAI is hampered by several critical barriers, such as data privacy, ethical risks, worker resistance, quality control issues, and infrastructure constraints. This research pinpoints these essential challenges and offers practical solutions. It provides actionable insights for businesses seeking to leverage GAI for competitive advantage in the evolving digital landscape by bridging the gap between theory and practice. The findings contribute to the growing discourse on AI-driven marketing strategies and lay the foundation for future research on GAI's long-term impact on consumer engagement and brand loyalty.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100630"},"PeriodicalIF":4.5,"publicationDate":"2025-12-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921252","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

ZKNiS-PoW: A privacy-preserving proof of ownership scheme for secure cloud storage ZKNiS-PoW：用于安全云存储的隐私保护所有权证明方案

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-03 DOI: 10.1016/j.array.2025.100627

Tang Zhou , Le Wang , Minxian Liang

There is a large amount of redundant data among users of cloud storage services. Client-side deduplication helps reduce the cost for service providers by avoiding repeated uploads and storage. However, this technique brings new security risks. Malicious users may use illegally obtained deduplication tags, such as file fingerprints, to fake ownership of other users’ files. Proof of Ownership (PoW) can require users to prove they have the full file, but existing methods are inefficient. They often need multiple rounds of interaction or complex computation over the whole file. As a result, the verification time increases with file size. To solve this problem, we propose a non-interactive PoW scheme based on zk-STARK. The system selects a number of challenge blocks that meet cryptographic security. It uses arithmetic circuits to encode block selection, hash computation, and the correctness of accumulators. Users only need to generate a zero-knowledge proof on these blocks. This allows them to prove they own the full file without revealing its content. The verification time does not depend on file size and appears near-constant in practice. In tests on files from 64 MB to 1 GB, our scheme is 1.2 to 46 times faster than existing methods. Security analysis shows that only a small number of blocks need to be verified. Even if an attacker knows 90% of the file, the chance of forgery is still lower than

2^{- 80}

. This scheme provides an efficient and practical solution for deduplication in cloud storage with strong privacy protection.

云存储服务用户之间存在大量冗余数据。客户端重复数据删除可以避免重复上传和存储，从而帮助服务提供商降低成本。然而，这种技术也带来了新的安全风险。恶意用户可能会利用非法获取的重复数据删除标签（如文件指纹）来假冒其他用户的文件所有权。所有权证明（PoW）可以要求用户证明他们拥有完整的文件，但是现有的方法效率很低。它们通常需要对整个文件进行多轮交互或复杂的计算。因此，验证时间随着文件大小的增加而增加。为了解决这一问题，我们提出了一种基于zk-STARK的非交互式PoW方案。系统选择一定数量满足加密安全性的挑战块。它使用算术电路对块选择、哈希计算和累加器的正确性进行编码。用户只需要在这些区块上生成零知识证明。这允许他们证明他们拥有完整的文件而不泄露其内容。验证时间不依赖于文件大小，并且在实践中几乎是恒定的。在对从64 MB到1 GB的文件进行测试时，我们的方案比现有方法快1.2到46倍。安全性分析表明，只需要验证少量块。即使攻击者知道90%的文件，伪造的几率仍然小于2−80。该方案为云存储的重复数据删除提供了一种高效实用的解决方案，同时具有较强的隐私保护能力。

{"title":"ZKNiS-PoW: A privacy-preserving proof of ownership scheme for secure cloud storage","authors":"Tang Zhou , Le Wang , Minxian Liang","doi":"10.1016/j.array.2025.100627","DOIUrl":"10.1016/j.array.2025.100627","url":null,"abstract":"<div><div>There is a large amount of redundant data among users of cloud storage services. Client-side deduplication helps reduce the cost for service providers by avoiding repeated uploads and storage. However, this technique brings new security risks. Malicious users may use illegally obtained deduplication tags, such as file fingerprints, to fake ownership of other users’ files. Proof of Ownership (PoW) can require users to prove they have the full file, but existing methods are inefficient. They often need multiple rounds of interaction or complex computation over the whole file. As a result, the verification time increases with file size. To solve this problem, we propose a non-interactive PoW scheme based on zk-STARK. The system selects a number of challenge blocks that meet cryptographic security. It uses arithmetic circuits to encode block selection, hash computation, and the correctness of accumulators. Users only need to generate a zero-knowledge proof on these blocks. This allows them to prove they own the full file without revealing its content. The verification time does not depend on file size and appears near-constant in practice. In tests on files from 64 MB to 1 GB, our scheme is 1.2 to 46 times faster than existing methods. Security analysis shows that only a small number of blocks need to be verified. Even if an attacker knows 90% of the file, the chance of forgery is still lower than <span><math><msup><mrow><mn>2</mn></mrow><mrow><mo>−</mo><mn>80</mn></mrow></msup></math></span>. This scheme provides an efficient and practical solution for deduplication in cloud storage with strong privacy protection.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100627"},"PeriodicalIF":4.5,"publicationDate":"2025-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145788380","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

IoT node for monitoring and traceability of live plants in maritime transport 用于监测和追溯海上运输中活植物的物联网节点

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-03 DOI: 10.1016/j.array.2025.100621

Blanca Méndez, Paula Lamo

The work presents an Internet of Things (IoT) monitoring node designed to convert shipping containers into smart greenhouses, thereby optimizing the transportation of live plants over long distances. This node utilizes specialized sensors to measure ambient temperature, soil temperature, relative humidity, soil moisture, and luminosity. Communication is via LoRa for local transmission and a low-Earth orbit (LEO) satellite infrastructure to ensure global connectivity. The collected data is stored and verified on the IOTA Tangle, ensuring traceability and immutability. Field tests in controlled environments demonstrated measurement accuracy, operational stability under varying environmental conditions, and optimized energy consumption. Additionally, the system is accessible through the Blynk platform, which provides real-time monitoring and customizable alert configurations. The paper also analyzes system costs, scalability, and specific conditions for installation on ships and containers. Limitations, such as dependence on satellite connectivity and structural barriers in maritime environments, are discussed. Finally, future lines of research focused on integrating artificial intelligence, energy optimization and connection with global logistics chains are identified.

该作品展示了一个物联网（IoT）监控节点，旨在将集装箱转换为智能温室，从而优化活植物的长距离运输。该节点利用专门的传感器测量环境温度、土壤温度、相对湿度、土壤湿度和亮度。通信通过LoRa进行本地传输，并通过低地球轨道（LEO）卫星基础设施确保全球连接。收集到的数据在IOTA Tangle上存储和验证，确保可追溯性和不变性。在受控环境下的现场测试证明了测量的准确性、在不同环境条件下的运行稳定性和优化的能耗。此外，该系统可通过Blynk平台访问，该平台提供实时监控和可定制的警报配置。本文还分析了该系统的成本、可扩展性以及在船舶和集装箱上安装的具体条件。讨论了对卫星连通性的依赖和海洋环境中的结构障碍等限制。最后，确定了未来的研究方向，重点是整合人工智能，能源优化和与全球物流链的连接。

{"title":"IoT node for monitoring and traceability of live plants in maritime transport","authors":"Blanca Méndez, Paula Lamo","doi":"10.1016/j.array.2025.100621","DOIUrl":"10.1016/j.array.2025.100621","url":null,"abstract":"<div><div>The work presents an Internet of Things (IoT) monitoring node designed to convert shipping containers into smart greenhouses, thereby optimizing the transportation of live plants over long distances. This node utilizes specialized sensors to measure ambient temperature, soil temperature, relative humidity, soil moisture, and luminosity. Communication is via LoRa for local transmission and a low-Earth orbit (LEO) satellite infrastructure to ensure global connectivity. The collected data is stored and verified on the IOTA Tangle, ensuring traceability and immutability. Field tests in controlled environments demonstrated measurement accuracy, operational stability under varying environmental conditions, and optimized energy consumption. Additionally, the system is accessible through the Blynk platform, which provides real-time monitoring and customizable alert configurations. The paper also analyzes system costs, scalability, and specific conditions for installation on ships and containers. Limitations, such as dependence on satellite connectivity and structural barriers in maritime environments, are discussed. Finally, future lines of research focused on integrating artificial intelligence, energy optimization and connection with global logistics chains are identified.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100621"},"PeriodicalIF":4.5,"publicationDate":"2025-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921256","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Design of a GenAI UX layer with small language models for edge computing in smart agriculture 基于小语言模型的智能农业边缘计算GenAI UX层设计

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-03 DOI: 10.1016/j.array.2025.100632

Juan M. Núñez V. , Carlos Alberto Peláez , Andrés Solano , Juan M. Corchado , Fernando De la Prieta

This study proposes a framework for designing a GenAI-UX layer in edge computing systems, integrating user experience (UX) principles and small language models (SLM) to optimize human–machine interaction in agricultural environments. Applied to indoor coriander cultivation, the enhanced UX design reduced the harvest cycle from 45 to 32 days by optimizing irrigation and environmental conditions through a voice-controlled interface with audio generation. The implementation of a structured UX approach enabled more efficient interaction with the system, facilitating data-driven decision-making processed at the edge. The study highlights the importance of UX design guidelines aligned with system requirements to ensure accessibility, intuitiveness, and efficiency in embedded environments. The results demonstrate that using SLM in edge systems improves personalization and response speed, enhancing agricultural productivity. This work underscores the need for a reference framework for UX design in embedded systems, ensuring effective interactions and a greater impact on the efficiency of controlled crop environments.

本研究提出了在边缘计算系统中设计GenAI-UX层的框架，整合用户体验（UX）原则和小语言模型（SLM）来优化农业环境中的人机交互。将增强的UX设计应用于室内香菜种植，通过具有音频生成功能的声控界面优化灌溉和环境条件，将收获周期从45天减少到32天。结构化用户体验方法的实现使与系统的交互更有效，促进了在边缘处理数据驱动的决策。该研究强调了用户体验设计指南与系统需求保持一致的重要性，以确保嵌入式环境中的可访问性、直观性和效率。结果表明，在边缘系统中使用SLM可以提高个性化和响应速度，从而提高农业生产力。这项工作强调了嵌入式系统中UX设计的参考框架的必要性，以确保有效的交互和对受控作物环境效率的更大影响。

{"title":"Design of a GenAI UX layer with small language models for edge computing in smart agriculture","authors":"Juan M. Núñez V. , Carlos Alberto Peláez , Andrés Solano , Juan M. Corchado , Fernando De la Prieta","doi":"10.1016/j.array.2025.100632","DOIUrl":"10.1016/j.array.2025.100632","url":null,"abstract":"<div><div>This study proposes a framework for designing a GenAI-UX layer in edge computing systems, integrating user experience (UX) principles and small language models (SLM) to optimize human–machine interaction in agricultural environments. Applied to indoor coriander cultivation, the enhanced UX design reduced the harvest cycle from 45 to 32 days by optimizing irrigation and environmental conditions through a voice-controlled interface with audio generation. The implementation of a structured UX approach enabled more efficient interaction with the system, facilitating data-driven decision-making processed at the edge. The study highlights the importance of UX design guidelines aligned with system requirements to ensure accessibility, intuitiveness, and efficiency in embedded environments. The results demonstrate that using SLM in edge systems improves personalization and response speed, enhancing agricultural productivity. This work underscores the need for a reference framework for UX design in embedded systems, ensuring effective interactions and a greater impact on the efficiency of controlled crop environments.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100632"},"PeriodicalIF":4.5,"publicationDate":"2025-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145684010","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Efficient crowd anomaly detection using C3D-LSTM networks with enhanced attention mechanisms 基于增强关注机制的C3D-LSTM网络的高效人群异常检测

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-03 DOI: 10.1016/j.array.2025.100625

Sarah Altowairqi , Suhuai Luo , Peter Greer , Shan Chen

The rising deployment of surveillance systems in urban environments necessitates efficient automated anomaly detection methods. While showing promise, current deep learning approaches struggle with computational complexity and real-time performance in processing spatiotemporal information. This paper presents a hybrid framework integrating Convolutional 3D Networks (C3D), Long Short-Term Memory (LSTM) networks, and attention mechanisms for anomaly detection. Through a systematic evaluation of four attention mechanisms—self-attention, multi-head attention, Bahdanau attention, and Luong attention—we demonstrate their operational differences and their differential impact on feature extraction and classification performance across three diverse benchmark datasets. Our multi-head attention variant achieves state-of-the-art results with 99.40 % accuracy and 99.96 % Area Under the Curve (AUC) on Violent Flows, while maintaining robust performance across varying dataset complexities, achieving 91.87 % accuracy on the ShanghaiTech Campus and 79.7 % accuracy on the UCF-Crime dataset. Comprehensive cross-dataset evaluation demonstrates consistent improvements of 2.4 %–3.5 % over baseline approaches, with all attention mechanisms outperforming traditional spatiotemporal models. The proposed architecture effectively balances computational requirements with detection performance, maintaining real-time processing capabilities suitable for operational deployment. This framework advances the technical capabilities of anomaly detection systems while providing a validated foundation for practical deployment in diverse surveillance environments, from controlled scenarios to challenging real-world conditions.

越来越多的监控系统部署在城市环境中，需要高效的自动化异常检测方法。虽然显示出前景，但目前的深度学习方法在处理时空信息方面存在计算复杂性和实时性的问题。本文提出了一种将卷积三维网络（C3D）、长短期记忆（LSTM）网络和注意机制集成在一起的混合框架，用于异常检测。通过对四种注意机制（自我注意、多头注意、巴赫达瑙注意和隆注意）的系统评估，我们展示了它们在三个不同基准数据集上的操作差异及其对特征提取和分类性能的差异影响。我们的多头注意力变体在暴力流上实现了最先进的结果，准确率为99.40%，曲线下面积（AUC）为99.96%，同时在不同的数据集复杂性下保持稳健的性能，在上海科技大学校园实现了91.87%的准确率，在ucf -犯罪数据集上实现了79.7%的准确率。综合跨数据集评估表明，与基线方法相比，该方法的一致性改进为2.4% - 3.5%，所有注意力机制都优于传统的时空模型。所提出的体系结构有效地平衡了计算需求和检测性能，保持了适合作战部署的实时处理能力。该框架提高了异常检测系统的技术能力，同时为各种监控环境（从受控场景到具有挑战性的现实条件）的实际部署提供了有效的基础。

{"title":"Efficient crowd anomaly detection using C3D-LSTM networks with enhanced attention mechanisms","authors":"Sarah Altowairqi , Suhuai Luo , Peter Greer , Shan Chen","doi":"10.1016/j.array.2025.100625","DOIUrl":"10.1016/j.array.2025.100625","url":null,"abstract":"<div><div>The rising deployment of surveillance systems in urban environments necessitates efficient automated anomaly detection methods. While showing promise, current deep learning approaches struggle with computational complexity and real-time performance in processing spatiotemporal information. This paper presents a hybrid framework integrating Convolutional 3D Networks (C3D), Long Short-Term Memory (LSTM) networks, and attention mechanisms for anomaly detection. Through a systematic evaluation of four attention mechanisms—self-attention, multi-head attention, Bahdanau attention, and Luong attention—we demonstrate their operational differences and their differential impact on feature extraction and classification performance across three diverse benchmark datasets. Our multi-head attention variant achieves state-of-the-art results with 99.40 % accuracy and 99.96 % Area Under the Curve (AUC) on Violent Flows, while maintaining robust performance across varying dataset complexities, achieving 91.87 % accuracy on the ShanghaiTech Campus and 79.7 % accuracy on the UCF-Crime dataset. Comprehensive cross-dataset evaluation demonstrates consistent improvements of 2.4 %–3.5 % over baseline approaches, with all attention mechanisms outperforming traditional spatiotemporal models. The proposed architecture effectively balances computational requirements with detection performance, maintaining real-time processing capabilities suitable for operational deployment. This framework advances the technical capabilities of anomaly detection systems while providing a validated foundation for practical deployment in diverse surveillance environments, from controlled scenarios to challenging real-world conditions.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100625"},"PeriodicalIF":4.5,"publicationDate":"2025-12-03","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145684012","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A survey of lightweight methods for object detection networks 目标检测网络的轻量级方法综述

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-02 DOI: 10.1016/j.array.2025.100589

Jing He, Jianfei Jiang, Changfan Zhang

As social production technologies develop, object detection becomes vital in sectors such as agriculture, industry, and healthcare. It decreases dependence on manual labour and enhances accuracy and efficiency. However, edge devices confront limitations in computational power, storage, and energy, creating a trade-off between accuracy and model size. To tackle this, academia and industry have proposed solutions including hardware-coordinated acceleration, adaptive task lightweighting, and hybrid compression. This paper reviews research from 2020 to 2025 on lightweight object detection, providing a systematic overview of efficient architecture and model compression techniques, explaining their mechanisms, challenges, and future directions to support ongoing progress.

随着社会生产技术的发展，目标检测在农业、工业和医疗保健等部门变得至关重要。它减少了对体力劳动的依赖，提高了准确性和效率。然而，边缘设备面临着计算能力、存储和能量方面的限制，需要在精度和模型尺寸之间进行权衡。为了解决这个问题，学术界和工业界提出了包括硬件协调加速、自适应任务轻量化和混合压缩在内的解决方案。本文回顾了从2020年到2025年在轻量目标检测方面的研究，提供了高效架构和模型压缩技术的系统概述，解释了它们的机制、挑战和未来方向，以支持正在进行的进展。

引用次数: 0

A systematic literature review on deep learning approaches for small object detection 对小目标检测的深度学习方法进行了系统的文献综述

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-02 DOI: 10.1016/j.array.2025.100615

Javed Sayyad, Khush Attarde

Object detection is essential in several industries, including defense, autonomous vehicles, and surveillance. These applications rely on various devices equipped with cameras, such as vehicles, drones, and satellites; primarily operating in the visible spectral domain rather than infrared or other spectral ranges. Deep Learning (DL) techniques have significantly advanced the field of object detection, enabling the identification of various objects. However, detecting tiny objects remains a challenging task. Despite its difficulty, identifying small objects in images captured by these devices in the visible spectrum is crucial. It is essential to explore hybrid techniques and modifications in feature architectures to address the challenge of detecting tiny objects. Simple architectures often fall short in this regard, necessitating more sophisticated approaches. This paper systematically reviews different DL-based approaches researchers have previously employed to tackle this issue. A systematic literature review on SOD and DL techniques uses the ”Preferred Reporting Items for Systematic Reviews and Meta-Analysis” (PRISMA) methodology. It discusses various DL-based theoretical frameworks, including Reinforcement Learning and Generative Adversarial Networks, specifically for Small Object Detection (SOD) in visible spectral images. The review begins by defining a small object and identifying the datasets available for various applications, such as remote sensing and autonomous vehicles. It then examines the implementation of models according to these datasets and analyzes the findings from other researchers. The analysis reveals that, for most datasets, the average precision (AP) for SOD ranges from 20% to 40% and showcases the need for the advancement and focus.

物体检测在国防、自动驾驶汽车和监视等多个行业都是必不可少的。这些应用依赖于配备摄像头的各种设备，如车辆、无人机和卫星；主要在可见光谱范围内工作，而不是在红外或其他光谱范围内工作。深度学习（DL）技术极大地推动了物体检测领域的发展，使识别各种物体成为可能。然而，探测微小物体仍然是一项具有挑战性的任务。尽管困难重重，但在这些设备捕获的可见光谱图像中识别小物体是至关重要的。探索混合技术和特征架构的修改来解决检测微小物体的挑战是至关重要的。简单的体系结构往往在这方面做得不够，需要更复杂的方法。本文系统地回顾了研究人员以前用来解决这个问题的不同的基于dl的方法。对SOD和DL技术的系统文献综述使用了“系统评价和荟萃分析的首选报告项目”（PRISMA）方法。它讨论了各种基于dl的理论框架，包括强化学习和生成对抗网络，特别是用于可见光谱图像中的小目标检测（SOD）。审查首先定义一个小对象，并确定可用于各种应用的数据集，如遥感和自动驾驶汽车。然后根据这些数据集检查模型的实现，并分析其他研究人员的发现。分析表明，对于大多数数据集，SOD的平均精度（AP）在20%到40%之间，这表明需要改进和关注。

{"title":"A systematic literature review on deep learning approaches for small object detection","authors":"Javed Sayyad, Khush Attarde","doi":"10.1016/j.array.2025.100615","DOIUrl":"10.1016/j.array.2025.100615","url":null,"abstract":"<div><div>Object detection is essential in several industries, including defense, autonomous vehicles, and surveillance. These applications rely on various devices equipped with cameras, such as vehicles, drones, and satellites; primarily operating in the visible spectral domain rather than infrared or other spectral ranges. Deep Learning (DL) techniques have significantly advanced the field of object detection, enabling the identification of various objects. However, detecting tiny objects remains a challenging task. Despite its difficulty, identifying small objects in images captured by these devices in the visible spectrum is crucial. It is essential to explore hybrid techniques and modifications in feature architectures to address the challenge of detecting tiny objects. Simple architectures often fall short in this regard, necessitating more sophisticated approaches. This paper systematically reviews different DL-based approaches researchers have previously employed to tackle this issue. A systematic literature review on SOD and DL techniques uses the ”Preferred Reporting Items for Systematic Reviews and Meta-Analysis” (PRISMA) methodology. It discusses various DL-based theoretical frameworks, including Reinforcement Learning and Generative Adversarial Networks, specifically for Small Object Detection (SOD) in visible spectral images. The review begins by defining a small object and identifying the datasets available for various applications, such as remote sensing and autonomous vehicles. It then examines the implementation of models according to these datasets and analyzes the findings from other researchers. The analysis reveals that, for most datasets, the average precision (AP) for SOD ranges from 20% to 40% and showcases the need for the advancement and focus.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"29 ","pages":"Article 100615"},"PeriodicalIF":4.5,"publicationDate":"2025-12-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145921199","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A quantitative approach to modeling transcription factor binding specificity from DNA sequences 从DNA序列建模转录因子结合特异性的定量方法

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-01 DOI: 10.1016/j.array.2025.100607

Yonglin Zhang , Shiqi Wu , Runyu Jing , Jiesi Luo

Understanding the mechanisms governing transcription factor (TF) binding specificity remains a fundamental challenge in computational biology. This study aims to improve the prediction of TF binding specificity, addressing key limitations in current computational models that often struggle to accurately predict TF binding affinities due to issues such as batch effects and non-uniform data distributions. Here, we present a novel approach that utilizes regression models trained on high-throughput SELEX (HT-SELEX) data, employing a deep forest algorithm to model the binding specificity of 215 mammalian TFs, integrating DNA sequences, composition, physicochemical properties, and structural features. The predicted binding affinities show strong concordance with experimental HT-SELEX measurements, confirming the model's accuracy and predictive reliability. Additionally, we apply this method to analyze the effects of genetic variations on TF binding, as well as to identify TF binding sites across the genome. Both applications rely on the same underlying methodology, with cumulative density function-based regression calibration improving the precision of predictions. This ensures reliable predictions of TF binding changes, both for sequence variations and genome-wide binding patterns. This study presents a novel approach that enhances the accuracy of TF binding specificity predictions and provides deeper insights into gene regulation and the impact of genetic variations on TF binding, with important implications for understanding gene regulation and disease mechanisms.

理解转录因子（TF）结合特异性的调控机制仍然是计算生物学的一个基本挑战。本研究旨在改进对TF结合特异性的预测，解决当前计算模型中由于批次效应和数据分布不均匀等问题而难以准确预测TF结合亲和力的关键限制。在这里，我们提出了一种新的方法，利用高通量SELEX （HT-SELEX）数据训练的回归模型，采用深度森林算法来模拟215种哺乳动物tf的结合特异性，整合DNA序列，组成，物理化学性质和结构特征。预测的结合亲和力与实验HT-SELEX测量结果具有较强的一致性，证实了模型的准确性和预测的可靠性。此外，我们应用该方法分析遗传变异对TF结合的影响，并确定整个基因组中的TF结合位点。这两种应用程序都依赖于相同的基本方法，基于累积密度函数的回归校准提高了预测的精度。这确保了对序列变异和全基因组结合模式的TF结合变化的可靠预测。本研究提出了一种新的方法，提高了TF结合特异性预测的准确性，并为基因调控和遗传变异对TF结合的影响提供了更深入的见解，对理解基因调控和疾病机制具有重要意义。

{"title":"A quantitative approach to modeling transcription factor binding specificity from DNA sequences","authors":"Yonglin Zhang , Shiqi Wu , Runyu Jing , Jiesi Luo","doi":"10.1016/j.array.2025.100607","DOIUrl":"10.1016/j.array.2025.100607","url":null,"abstract":"<div><div>Understanding the mechanisms governing transcription factor (TF) binding specificity remains a fundamental challenge in computational biology. This study aims to improve the prediction of TF binding specificity, addressing key limitations in current computational models that often struggle to accurately predict TF binding affinities due to issues such as batch effects and non-uniform data distributions. Here, we present a novel approach that utilizes regression models trained on high-throughput SELEX (HT-SELEX) data, employing a deep forest algorithm to model the binding specificity of 215 mammalian TFs, integrating DNA sequences, composition, physicochemical properties, and structural features. The predicted binding affinities show strong concordance with experimental HT-SELEX measurements, confirming the model's accuracy and predictive reliability. Additionally, we apply this method to analyze the effects of genetic variations on TF binding, as well as to identify TF binding sites across the genome. Both applications rely on the same underlying methodology, with cumulative density function-based regression calibration improving the precision of predictions. This ensures reliable predictions of TF binding changes, both for sequence variations and genome-wide binding patterns. This study presents a novel approach that enhances the accuracy of TF binding specificity predictions and provides deeper insights into gene regulation and the impact of genetic variations on TF binding, with important implications for understanding gene regulation and disease mechanisms.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100607"},"PeriodicalIF":4.5,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145614679","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

A real-valued DCT-based spectral CNN architecture for efficient edge deep learning 一种基于实值dct的高效边缘深度学习频谱CNN架构

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-01 DOI: 10.1016/j.array.2025.100594

Ibrahim Yousef Alshareef , Ab Al-Hadi Ab Rahman , Mohamed Khalafalla Hassan , Mohd Shahrizal Rusli , Nuzhat Khan , Ali Manzak , Muhammad Nadzir Marsono

Spectral Convolutional Neural Networks (SpCNNs) offer a pathway to computational efficiency by performing convolutions in the frequency domain. While FFT-based SpCNNs reduce convolutional complexity, their reliance on complex valued operations and inverse transforms incurs high memory and latency costs, limiting their utility for embedded and edge applications. This paper proposes an enhanced, real-valued Discrete Cosine Transform (DCT)-based SpCNN that eliminates inverse transforms and complex arithmetic by performing all operations, including convolution, activation, and pooling entirely in the DCT domain. A modified real-valued spectral activation function is introduced to enable effective nonlinearity in frequency space. Experimental evaluation on MNIST and a 94-class ASCII benchmark datasets demonstrates that the proposed model achieves up to 20% reduction in computational workload and 19% reduction in memory access compared to a previous spectral model. Additionally, LeNet5 achieves improved accuracy (98.44%), and the architecture exhibits significantly faster inference and higher energy efficiency in both batch and real time settings. These results establish the DCT-based SpCNN as a practical and scalable solution for deployment in resource constrained systems.

频谱卷积神经网络（SpCNNs）通过在频域进行卷积提供了一种提高计算效率的途径。虽然基于fft的spcnn降低了卷积复杂度，但它们对复杂值运算和逆变换的依赖导致了高内存和延迟成本，限制了它们在嵌入式和边缘应用中的实用性。本文提出了一种增强的、基于实值离散余弦变换（DCT）的SpCNN，它通过完全在DCT域中执行包括卷积、激活和池化在内的所有操作来消除逆变换和复杂的算法。引入了一种改进的实值谱激活函数来实现频率空间的有效非线性。在MNIST和94类ASCII基准数据集上的实验评估表明，与之前的频谱模型相比，该模型的计算工作量减少了20%，内存访问减少了19%。此外，LeNet5实现了更高的准确率（98.44%），并且该架构在批处理和实时设置中都表现出更快的推理和更高的能源效率。这些结果表明，基于dct的SpCNN是一种在资源受限系统中部署的实用且可扩展的解决方案。

{"title":"A real-valued DCT-based spectral CNN architecture for efficient edge deep learning","authors":"Ibrahim Yousef Alshareef , Ab Al-Hadi Ab Rahman , Mohamed Khalafalla Hassan , Mohd Shahrizal Rusli , Nuzhat Khan , Ali Manzak , Muhammad Nadzir Marsono","doi":"10.1016/j.array.2025.100594","DOIUrl":"10.1016/j.array.2025.100594","url":null,"abstract":"<div><div>Spectral Convolutional Neural Networks (SpCNNs) offer a pathway to computational efficiency by performing convolutions in the frequency domain. While FFT-based SpCNNs reduce convolutional complexity, their reliance on complex valued operations and inverse transforms incurs high memory and latency costs, limiting their utility for embedded and edge applications. This paper proposes an enhanced, real-valued Discrete Cosine Transform (DCT)-based SpCNN that eliminates inverse transforms and complex arithmetic by performing all operations, including convolution, activation, and pooling entirely in the DCT domain. A modified real-valued spectral activation function is introduced to enable effective nonlinearity in frequency space. Experimental evaluation on MNIST and a 94-class ASCII benchmark datasets demonstrates that the proposed model achieves up to 20% reduction in computational workload and 19% reduction in memory access compared to a previous spectral model. Additionally, LeNet5 achieves improved accuracy (98.44%), and the architecture exhibits significantly faster inference and higher energy efficiency in both batch and real time settings. These results establish the DCT-based SpCNN as a practical and scalable solution for deployment in resource constrained systems.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100594"},"PeriodicalIF":4.5,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145614762","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Anomaly detection in industrial control systems: Leveraging adaptive multi-granularity anomaly correction for robust performance in noisy environments 工业控制系统中的异常检测：利用自适应多粒度异常校正在噪声环境中的鲁棒性能

IF 4.5 Q2 COMPUTER SCIENCE, THEORY & METHODS

Array

Pub Date : 2025-12-01 DOI: 10.1016/j.array.2025.100588

Yaofang Zhang , Sicai Lv , Yang Liu , Hongri Liu , Bailing Wang

Anomaly detection in industrial control systems (ICS) is critical for ensuring process reliability and protecting human life and property. Prediction-deviation-based models have attracted significant attention because they require minimal labeled anomaly data. However, prior studies have mainly focused on improving prediction accuracy, thereby increasing sensitivity to data deviations. These data deviation-based methods typically treat each time step as an individual detection unit, overlooking the interdependencies within the context of abnormal states. As a result, oscillating anomaly states are often produced, which significantly lower detection accuracy. Moreover, the uncertainty of industrial environments can introduce substantial noise or cause data shifts, posing additional challenges for anomaly detection. To tackle these challenges in complex noisy environments, we propose a multivariate time series anomaly detection method with adaptive multi-granularity anomaly correction (ATLAC). By incorporating an attention mechanism, ATLAC assigns weights to frequency-decomposed sub-series, emphasizing key frequency components while suppressing noise. We apply the TCN-LSTM architecture to model the reconstructed sub-series, using data deviation distributions to preliminarily detect anomalies. Furthermore, to account for the contextual influence of abnormal states, we adaptively set thresholds based on state information within multi-granularity windows, correcting frequent oscillations while eliminating short-term noise interference. Experimental results on both raw and noise-enhanced Tennessee Eastman datasets demonstrate that our method outperforms baseline methods in terms of detection accuracy across various scenarios, especially in noisy environments, where ATLAC significantly reduces the false positive rate by up to 13.62%.

工业控制系统（ICS）中的异常检测对于确保过程可靠性和保护人类生命财产安全至关重要。基于预测偏差的模型已经引起了极大的关注，因为它们需要最小的标记异常数据。然而，以往的研究主要集中在提高预测精度，从而增加对数据偏差的敏感性。这些基于数据偏差的方法通常将每个时间步视为单个检测单元，忽略了异常状态上下文中的相互依赖性。因此，经常产生振荡异常状态，这大大降低了检测精度。此外，工业环境的不确定性可能会引入大量噪声或导致数据移位，这给异常检测带来了额外的挑战。为了解决复杂噪声环境下的这些问题，我们提出了一种自适应多粒度异常校正（ATLAC）的多变量时间序列异常检测方法。通过引入注意机制，ATLAC为频率分解子序列分配权重，在抑制噪声的同时强调关键频率分量。我们采用TCN-LSTM架构对重构子序列进行建模，利用数据偏差分布初步检测异常。此外，为了考虑异常状态的上下文影响，我们根据多粒度窗口内的状态信息自适应设置阈值，在消除短期噪声干扰的同时纠正频繁的振荡。在原始和噪声增强的田纳西伊士曼数据集上的实验结果表明，我们的方法在各种场景下的检测精度优于基线方法，特别是在噪声环境中，ATLAC显著降低了假阳性率，最高可达13.62%。

{"title":"Anomaly detection in industrial control systems: Leveraging adaptive multi-granularity anomaly correction for robust performance in noisy environments","authors":"Yaofang Zhang , Sicai Lv , Yang Liu , Hongri Liu , Bailing Wang","doi":"10.1016/j.array.2025.100588","DOIUrl":"10.1016/j.array.2025.100588","url":null,"abstract":"<div><div>Anomaly detection in industrial control systems (ICS) is critical for ensuring process reliability and protecting human life and property. Prediction-deviation-based models have attracted significant attention because they require minimal labeled anomaly data. However, prior studies have mainly focused on improving prediction accuracy, thereby increasing sensitivity to data deviations. These data deviation-based methods typically treat each time step as an individual detection unit, overlooking the interdependencies within the context of abnormal states. As a result, oscillating anomaly states are often produced, which significantly lower detection accuracy. Moreover, the uncertainty of industrial environments can introduce substantial noise or cause data shifts, posing additional challenges for anomaly detection. To tackle these challenges in complex noisy environments, we propose a multivariate time series anomaly detection method with adaptive multi-granularity anomaly correction (ATLAC). By incorporating an attention mechanism, ATLAC assigns weights to frequency-decomposed sub-series, emphasizing key frequency components while suppressing noise. We apply the TCN-LSTM architecture to model the reconstructed sub-series, using data deviation distributions to preliminarily detect anomalies. Furthermore, to account for the contextual influence of abnormal states, we adaptively set thresholds based on state information within multi-granularity windows, correcting frequent oscillations while eliminating short-term noise interference. Experimental results on both raw and noise-enhanced Tennessee Eastman datasets demonstrate that our method outperforms baseline methods in terms of detection accuracy across various scenarios, especially in noisy environments, where ATLAC significantly reduces the false positive rate by up to 13.62%.</div></div>","PeriodicalId":8417,"journal":{"name":"Array","volume":"28 ","pages":"Article 100588"},"PeriodicalIF":4.5,"publicationDate":"2025-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145614856","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

引用次数: 0

Array最新文献