·         Table of Contents (Volume #9, Issue #2)

·         Cover Page (Volume #9, Issue #2)

·         Editorial (Volume #9, Issue #2)

Paper #1

Efficient Object Detection in Remote Sensing Images Using Quantitative Augmentation and Competitive Learning

Huaxiang Song, Junping Xie, Yan Zhang, Yang Zhou, Wenhui Wang, YingYing Duan and Xinyi Xie

Abstract: Object detection in remote sensing images (RSIs) is crucial in Earth observation. However, current approaches often overlook key characteristics of RSIs, resulting in models that fail to balance accuracy and computational efficiency. To the authors’ knowledge, these limitations stem from the inherent scarcity and complexity of RSI samples, which cannot be fully resolved by solely modifying the model architecture. To address these challenges, we propose QACL-Net, an object detection method built on the Faster R-CNN framework, which significantly enhances the performance of CNN-based detectors for RSI recognition while maintaining fast inference speeds. QACL-Net incorporates several innovative techniques. Firstly, we introduce the quantitative augmentation (QA) strategy to address RSI sample scarcity. Secondly, we propose the equal-quadrate mosaic (EQM) algorithm to improve the effectiveness of the traditional mosaic technique for RSI detection. Thirdly, we implement the competitive learning (CL) strategy to resolve the problem of redundant feature fusion in the feature pyramid network. Crucially, the proposed enhancement techniques are integrated into three plug-and-play modules. To evaluate the proposed method, we develop two variants of QACL-Net by utilizing an EfficientNet-B0 and EfficientNet-B3 backbone model for the detector architecture, respectively. Extensive experiments on two widely used RSI datasets demonstrate that QACL-Net outperforms 31 advanced methods since 2022 on the DIOR20 dataset. Specifically, QACL-Net-B3 achieves a 6.9% improvement in accuracy on the challenging DIOR20 dataset. Additionally, QACL-Net-B3 reduces model size by 33% and increases inference speed by 17% compared to the baseline model. In summary, our work highlights the significant impact of RSI sample scarcity, noisy backgrounds, and feature fusion redundancy on object detection performance. Theoretically, our approach can be seamlessly integrated with other detection models, as the QA, EQM, and CL modules require only minimal modifications to the model structure.

Keywords: Competitive Learning; Equal-Quadrate Mosaic; QACL-Net; Quantitative Augmentation; Remote Sensing Object Detection.

Download Full Text

Paper #2

Integration Named Entity Recognition and Latent Dirichlet Allocation to Enhance Topic Modeling

Hawraa Ali Taher, Noralhuda N. Alabid and Bushra Mahdi Hasan

Abstract: Topic modeling from texts is one of the important topics in natural language processing (NLP), as it plays a fundamental role in summarizing texts, understanding their content, and facilitating access to the main ideas, especially in light of the vast quantity of unstructured texts available today. Extracting titles is used in a variety of fields, such as news archiving, document classification, and content analysis in social media, making it an essential tool for improving information management and effective presentation. In this research, we focused on improving the methodology for extracting titles from texts by integrating two leading techniques: the topic assignment model using Latent Dirichlet Allocation (LDA) and the named entity recognition technique (NER). This combination aims to achieve a balance between identifying general topics of texts via LDA and extracting important information and key entities using NER, ensuring the generation of accurate and understandable titles that better reflect the actual content of the texts. The results of the study showed that the combined methodology achieved an accuracy of 71.97%, outperforming the performance of each technique separately, where the accuracy of NER alone was 29.78% and the accuracy of LDA alone was 67.80%. These results underscore the importance of integrating different techniques into NLP to improve headline extraction performance. This approach contributes to the development of more efficient text analysis methods, which enhances NLP applications in areas such as news analysis, content management, and document summarization, highlighting the importance of the topic in improving the handling of large texts and presenting them in a clearer and more appropriate way.

Keywords: Bert Model; Cosine Similarity; Latent Dirichlet Allocation (LDA); Name Entity Recognition (NER); Text Analysis; Topic Modeling.

Download Full Text

Paper #3

Blockchain Aided Smart Consensus Model for IoMT Architecture

Md. Iftekharul Alam Efat, Tasnim Rahman, Md. Jane Alam and Shah Mostafa Khaled

Abstract: The rapid evolution of technology in healthcare underscores its pivotal role in shaping human lives, particularly with the widespread adoption of wearable Internet of Things (IoT) sensors. This surge has led to the interconnection of a vast array of devices, sensors, and real-time trackers over the internet, driving research interest in the development of a Remote Patient Monitoring system for treatment and consultation. Concurrently, the integration of Artificial Intelligence (AI) into decision support systems has become prevalent, paving the way for the creation of a smart consensus system that enhances efficiency through live remote sensor data. Nevertheless, the adoption of such advanced technologies is not without challenges, particularly in the realms of privacy, security, and standardization for data transmission and transaction cataloguing. To address these concerns, blockchain has emerged as a compelling solution, seamlessly integrating with existing solutions and providing heightened security and privacy assurances, especially in collaboration with third-party entities. In response to these challenges, a proposed blockchain-enabled Internet of Medical Things (IoMT) architecture takes centre stage. This innovative framework leverages real-time wearable sensor data from patients as the primary source for diagnosis, ensuring secure storage and transmission through the application of blockchain technology. Furthermore, the proposed IoMT architecture extends its reach by integrating stakeholders from the healthcare domain. This collaboration enhances the system's effectiveness, with an AI-based decision support system aiding consultants in remotely monitoring patients with ease. However, the simulated experiment demonstrates improved access control, authentication, scalability (150 TPS), low energy consumption (50kWh/year), and reduced transaction latency (200ms) compared to PoW, PoS, and PoA. In summary, the fusion of wearable IoT sensors, AI, and blockchain in this IoMT architecture not only addresses the challenges of privacy and security but also establishes a robust foundation for advancing electronic healthcare systems.

Keywords: Antenna directivity; Electromagnetics; Grounding in monopole antennas; Image theory; Internet of Things; ISM band; Radial monopole; 433 MHz.

Download Full Text