17628 research outputs found
Sort by
Phonetic Error Analysis Beyond Phone Error Rate
No full textIn this article, we analyse the performance of the TIMIT-based phone recognition systems beyond the overall phone error rate (PER) metric. We consider three broad phonetic classes (BPCs): {affricate, diphthong, fricative, nasal, plosive, semi-vowel, vowel, silence}, {consonant, vowel, silence} and {voiced, unvoiced, silence} and, calculate the contribution of each phonetic class in terms of the substitution, deletion, insertion and PER. Furthermore, for each BPC we investigate the following: evolution of PER during training, effect of noise (NTIMIT), importance of different spectral subbands (1, 2, 4, and 8 kHz), usefulness of bidirectional vs unidirectional sequential modelling, transfer learning from WSJ and regularisation via monophones. In addition, we construct a confusion matrix for each BPC and analyse the confusions via dimensionality reduction to 2D at the input (acoustic features) and output (logits) levels of the acoustic model. We also compare the performance and confusion matrices of the BLSTM-based hybrid baseline system with those of the GMM-HMM based hybrid, Conformer and wav2vec 2.0 based end-to-end phone recognisers. Finally, the relationship of the unweighted and weighted PERs with the broad phonetic class priors is studied for both the hybrid and end-to-end systems
A novel autonomous container-based platform for cybersecurity training and research
Get PDFCyberattacks, particularly those targeting systems that store or handle sensitive data, have become more sophisticated in recent years. To face increasing threats, continuous capacity building and digital skill competence are needed. Cybersecurity hands-on training is essential to upskill cybersecurity professionals. However, the cost of developing and maintaining a cyber range platform is high. Setting up an ideal digital environment for cybersecurity exercises can be challenging and often need to invest a lot of time and system resources in this process. In this article, we present a lightweight cyber range platform that was developed under the open-source cloud platform OpenStack, based on Docker technology using IaC methodology. Combining the advantages of Docker technology, DevOps automation capabilities, and the cloud platform, the proposed cyber range platform achieves the maximization of performance and scalability while reducing costs and resources
Louis's Story: The Story of Appletree
No full textThe purpose of this project• To raise the visibility of people with profound and multiple learning disabilities and autism who as a marginalised group of people are imprisoned, abused and ignored along with their families.• To conceptualise loving justice and forgiveness in addressing systemic disablism.• To develop theological reflection and action as a new way of creating communities for profoundly vulnerable people within the context of social policy and its implementation.• To develop resources for empowering the informal sector and the building of intentional communities to address those systemic problems
An Order Fulfilment Location Planning Model for Perishable Goods Supply Chains Using Population Density
Get PDFThe increased frequency of purchases and growing distances between the final distribution points of the perishable products and consumers are contributing to multiple handling, high intermediation, and a greater concentration of outlets selling perishables goods. These in turn have increased logistics costs, and inefficient supply chain. This study presents a modelling framework for locating perishable goods order fulfilment centers (OFC) near the consumer by using population density as a proxy for demand. The centrality and Borda count measures are used to identify optimal locations in perishable goods supply chain networks. We present a case study to demonstrate the applicability and efficacy of the proposed density-based spatial methodology
FireXnet: an explainable AI-based tailored deep learning model for wildfire detection on resource-constrained devices
Get PDFBackground: Forests cover nearly one-third of the Earth’s land and are some of our most biodiverse ecosystems. Due to climate change, these essential habitats are endangered by increasing wildfires. Wildfires are not just a risk to the environment, but they also pose public health risks. Given these issues, there is an indispensable need for efficient and early detection methods. Conventional detection approaches fall short due to spatial limitations and manual feature engineering, which calls for the exploration and development of data-driven deep learning solutions. This paper, in this regard, proposes 'FireXnet', a tailored deep learning model designed for improved efficiency and accuracy in wildfire detection. FireXnet is tailored to have a lightweight architecture that exhibits high accuracy with significantly less training and testing time. It contains considerably reduced trainable and non-trainable parameters, which makes it suitable for resource-constrained devices. To make the FireXnet model visually explainable and trustable, a powerful explainable artificial intelligence (AI) tool, SHAP (SHapley Additive exPlanations) has been incorporated. It interprets FireXnet’s decisions by computing the contribution of each feature to the prediction. Furthermore, the performance of FireXnet is compared against five pre-trained models — VGG16, InceptionResNetV2, InceptionV3, DenseNet201, and MobileNetV2 — to benchmark its efficiency. For a fair comparison, transfer learning and fine-tuning have been applied to the aforementioned models to retrain the models on our dataset. Results: The test accuracy of the proposed FireXnet model is 98.42%, which is greater than all other models used for comparison. Furthermore, results of reliability parameters confirm the model’s reliability, i.e., a confidence interval of [0.97, 1.00] validates the certainty of the proposed model’s estimates and a Cohen’s kappa coefficient of 0.98 proves that decisions of FireXnet are in considerable accordance with the given data. Conclusion: The integration of the robust feature extraction of FireXnet with the transparency of explainable AI using SHAP enhances the model’s interpretability and allows for the identification of key characteristics triggering wildfire detections. Extensive experimentation reveals that in addition to being accurate, FireXnet has reduced computational complexity due to considerably fewer training and non-training parameters and has significantly fewer training and testing times
Breast Cancer Detection Based on Simplified Deep Learning Technique With Histopathological Image Using BreaKHis Database
Get PDFPresented here are the results of an investigation conducted to determine the effectiveness of deep learning (DL)‐based systems utilizing the power of transfer learning for detecting breast cancer in histopathological images. It is shown that DL models that are not specifically developed for breast cancer detection can be trained using transfer learning to effectively detect breast cancer in histopathological images. The outcome of the analysis enables the selection of the best DL architecture for detecting cancer with high accuracy. This should facilitate pathologists to achieve early diagnoses of breast cancer and administer appropriate treatment to the patient. The experimental work here used the BreaKHis database consisting of 7909 histopathological pictures from 82 clinical breast cancer patients. The strategy presented for DL training uses various image processing techniques for extracting various feature patterns. This is followed by applying transfer learning techniques in the deep convolutional networks like ResNet, ResNeXt, SENet, Dual Path Net, DenseNet, NASNet, and Wide ResNet. Comparison with recent literature shows that ResNext‐50, ResNext‐101, DPN131, DenseNet‐169 and NASNet‐A provide an accuracy of 99.8%, 99.5%, 99.675%, 99.725%, and 99.4%, respectively, and outperform previous studies
Industry 4.0 to Industry 5.0: Mapping the Transitions
No full textExplores the causes and consequences of transition from Industry 4.0 to Industry 5.0Highlights the human-centric approach of Industry 5.0Discusses Big Data, Artificial Intelligence, and Human–Robot coworking in Industry 5.
Information literacy and society: A report to present findings from a review of literature on the impact of information literacy on society
Get PDFThis report presents findings from a review of literature reporting on the impact of information literacy (IL) on society. It is intended to deliver considerations on how academic research into IL can positively affect society, building on the 2022 Information Literacy Impact Framework (ILIF) project report (https://napier-repository.worktribe.com/output/2910549/information-literacy-impact-framework-final-project-report).The project team has:•from a filtered set of over 4000 results, developed longlists corresponding to the five topics in the CILIP (2018) definition of IL, totalling 197 items for possible further review•noted research themes, barriers to IL research, barriers to and enablers of shaping information-literate populations emerging from the longlists•filtered the longlists to shortlists, totalling 35 items, for detailed review•classified the longlists and shortlists in two dimensions: geography and method of study•undertaken detailed analysis of the shortlist items.•drawn conclusions on the role of information literacy in society.The core research that investigates the role of IL in society is geographically skewed towards the anglosphere and the first world. The factors causing this skew are unclear, but extra apparent skew may have resulted from this project’s focus on English-language peer-reviewed publications. Education, particularly tertiary education, is significantly over-represented in the IL research literature. Barriers to shaping information-literate populations are raised by issues around IL teaching and structures that could support it, including government (in)action.Other key findings are:•IL research covers a very wide range of topics and contexts.•IL training/education should be delivered by collaboration between librarians and teachers/lecturers, continue throughout education, and be reinforced during careers and lifetimes.•IL research may have indirect impact, e.g. research into improving medical professionals’ IL does not just affect these professionals but also wider society, i.e. their patients.•There are missed opportunities for such societal impact, e.g. where medical professionals do not have IL skills and so may not give their patients the best treatment possible; if citizens do not have health information literacy their health may suffer.•Many of the findings from the ILIF project are validated
RLT 30 YEARS Research in Learning Technology: making friends and influencing people
Get PDFThe first issue of Research in Learning Technology (RLT) was published in 1993. Over 30 years, the journal has comprised an informal research and development facility for new ideas and practices in technology enhanced learning. This paper takes nine published articles from RLT: the three most downloaded in the. The aim is to identify different areas of current interest and influence, different areas of practice, and different scholarly approaches. The authors are the journal's current editorial team. This paper identifies diversity of technology enhanced learning-related subject matter and different approaches, too, but with ongoing interest in efficacy and in the 'how' of technology enhanced learning: how technology can be applied to truly enhance learning, comprising an approachable community, generating influence
Data sources:
Access Repository Dashboard
Do you manage Repository@Napier? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!