https://bura.brunel.ac.uk/handle/2438/13035 2026-11-05T13:52:19Z 2026-11-05T13:52:19Z Jiang, Feibo Dong, Li Pan, Cunhua Wang, Kezhi Huang, Yongming Yang, Kun Wang, Chengxiang Wang, Xianbin Jamalipour, Abbas Al-Dhahir, Naofal Wang, Jiangzhou You, Xiaohu https://bura.brunel.ac.uk/handle/2438/34070 2026-10-22T02:00:36Z 2026-01-01T00:00:00Z

Title: Toward Native Intelligence in 6G: A Comprehensive Survey of Large Models and Agentic AI for Communications Authors: Jiang, Feibo; Dong, Li; Pan, Cunhua; Wang, Kezhi; Huang, Yongming; Yang, Kun; Wang, Chengxiang; Wang, Xianbin; Jamalipour, Abbas; Al-Dhahir, Naofal; Wang, Jiangzhou; You, Xiaohu Abstract: ...

2026-01-01T00:00:00Z Ibrahim, Zear Money, Arthur Atwal, Anita Spiliotopoulou, georgia https://bura.brunel.ac.uk/handle/2438/33972 2026-10-10T02:00:38Z 2026-01-01T00:00:00Z

Title: OT-Vision: A 3D depth sensing mobile application to aid clinician-led home environment assessments for falls prevention Authors: Ibrahim, Zear; Money, Arthur; Atwal, Anita; Spiliotopoulou, georgia Abstract: ... Description: ...

2026-01-01T00:00:00Z Altarawneh, Ahmad Arzoky, Mahir Swift, Steven Al Fayez, Reem Qadan Al-Zoubi, Moh’d Belal Sowan, Bilal Zhang, Li https://bura.brunel.ac.uk/handle/2438/33899 2026-09-24T02:01:06Z 2025-12-29T00:00:00Z

Title: ETM-F: an enriched topic modeling and filtration framework integrating ontologies and deep learning for biomedical trend analysis Authors: Altarawneh, Ahmad; Arzoky, Mahir; Swift, Steven; Al Fayez, Reem Qadan; Al-Zoubi, Moh’d Belal; Sowan, Bilal; Zhang, Li Abstract: Selecting research topics from author keywords helps identify emerging trends in medicine, but keyword-only models remain fragile because of specialized terminology and rapidly evolving vocabularies. To address this limitation, we propose Enriched Topic Modeling with Filtration (ETM-F), a methodological framework that integrates frequency-based keyword filtration with Medical Subject Headings (MeSH) enrichment. ETM-F is model-agnostic and can be applied across probabilistic and neural topic models. We evaluate ETM-F against baseline Latent Dirichlet Allocation (LDA) and extend the analysis to hierarchical LDA (hLDA), Dynamic Topic Models (DTM), Contextualized Topic Models (CTM), and BERTopic. Model performance is assessed through perplexity, Cv coherence, normalized pointwise mutual information (NPMI), topic diversity, dominant topic contribution, and temporal coherence. Statistical significance is established through bootstrap confidence intervals and Wilcoxon signed-rank tests. Experiments on 232,191 medical publications from Scopus (2020) confirm the advantages of enrichment. For LDA, Cv coherence rises from 0.37 with author keywords to 0.52 with MeSH-enriched keywords, while NPMI improves from –0.05 to 0.08. Neural topic models provide higher quality results. After filtration, BERTopic achieves the best coherence (Cv = 0.75, NPMI = 0.28), followed by hierarchical LDA (Cv = 0.66) and CTM (Cv = 0.60). Dynamic Topic Models and time-sliced BERTopic capture temporal signals, including the shift from generic vaccine themes to mRNA-specific terms and the surge of telemedicine. Embedding-based keyword expansion with BioWordVec and BioBERT further enhances coherence by 0.03 to 0.05 and identifies synonyms such as “SARS-CoV-2”. These findings confirm that semantic enrichment and dynamic models improve topic discovery in large-scale biomedical text. ETM-F establishes a reproducible benchmark for future research and offers a reliable tool for identifying thematic directions in medical literature. Description: Data availability: No datasets were generated or analyzed during the current study.

2025-12-29T00:00:00Z Han, Juntao Wang, Zidong Liu, Weibo Song, Baoye Wang, Chuang Liu, Xiaohui https://bura.brunel.ac.uk/handle/2438/33855 2026-09-12T02:01:10Z 2026-08-07T00:00:00Z

Title: Data-centric deep learning for defect detection in fiber composite materials: A review of algorithms, challenges, and emerging trends Authors: Han, Juntao; Wang, Zidong; Liu, Weibo; Song, Baoye; Wang, Chuang; Liu, Xiaohui Abstract: Fiber composite materials (FCMs) have been widely used in high-performance engineering systems owing to their high strength-to-weight ratio, corrosion resistance, and excellent mechanical properties. Reliable defect detection is therefore essential for ensuring structural integrity, operational safety, and service-life extension. Deep learning (DL) has recently shown strong potential for automated FCMs defect detection by enabling powerful feature extraction, accurate localization, and robust representation learning from image data. Nevertheless, its performance remains highly dependent on sufficient, high-quality, and well-annotated training data, which are often difficult to obtain in practical detection scenarios. This paper presents a comprehensive review of image-based DL approaches for FCMs defect detection under different data conditions. Representative R-CNN-based, YOLO-based, and Transformer-based models are first reviewed for sufficient and high-quality data settings. DL-based denoising, data augmentation, and transfer learning strategies are then analyzed for limited or low-quality data settings, with emphasis on Autoencoders, Generative Adversarial Networks, Variational Autoencoders, Diffusion Models, and pretrained model adaptation. Key challenges and future directions are further discussed, including long-term structural health monitoring, physics-guided defect detection, explainable artificial intelligence, multimodal data fusion, and large language model-assisted detection. By organizing existing studies from a data-centric perspective, this review provides a structured reference for developing reliable, interpretable, and deployable DL-based defect detection systems for FCMs. Description: Data availability Data will be made available on request.

2026-08-07T00:00:00Z