17837 research outputs found
Sort by
“We Created Our Ecosystem”:A Multi-Country Case Study of Adoption Strategies for IoT in Agriculture
Get PDFIn this paper, we present preliminary findings from our ongoing research on the adoption of the Internet of Things (IoT) in agriculture, addressing the limited scholarly focus on practical implementation of IoT-based business innovations in the field. The study employs a multiple exploratory case study approach, focusing on four countries: Greece, Italy, Serbia, and Türkiye. Through qualitative methods, including in-depth interviews and field observations, we investigate how the diffusion of IoT innovations unfolds across diverse geographical and socio-economic contexts. Our findings underscore the early stage of the innovation ecosystem formation within the IoT market for agriculture, which is often fragmented, loosely coupled, and ephemeral. This necessitates that companies strategise effectively, aligning with the evolving and, in some cases, yet-to-emerge IoT ecosystem to ensure successful IoT-driven ventures. Our research extends the literature by proposing a new framework that offers guidance for practitioners in selecting suitable adoption strategies for IoT in agriculture
Dilemmatic commitments following psychological contract breach and its contrasting implications for voluntary turnover and organizational citizenship behaviour
Get PDF
A Tunable Depth Microwave Photonic Notch Filter Based on Polarization Control of Stimulated Brillouin Scattering
Get PDFIn this paper, we report a frequency tunable, suppression depth tunable, narrow Microwave Photonic notch filter based on the State of Polarisation (SOP) control of Stimulated Brillouin Scat-tering (SBS) in SMF. SBS gain is highly dependent on the SOPs of the pump and probe signals with a maximum gain when their SOPs are the same, and a minimum when they are orthogonal to one another. We exploit this phenomenon to create a tunable depth notch filter by adjusting the gain and loss of SBS for RF cancellation with a photodiode. We develop and present a math-ematical model explaining the polarisation dependent gain and the transfer function of the proposed filter. Experimental results demonstrate a tunable notch filter with a notch depth tuning range of 0-40dB with a frequency range of 2-26.5GHz. The -3dB width of the filter was recorded as 37MHz at its maximum depth, with a low 4MHz recorded in shallower notches. The notch depth is tuned by a polarisation controller that controls the SOP of the pump signal. This configuration has applications in dynamic notch filtering of signals such as Radio Frequency Interference (RFI) removal where control over the notch depth is of great importance. The proposed tunable depth notch filter would provide unique benefits for satellite, aerospace, telecom-munications and beyond communication technologies
Behavior change approaches for cyber security and the need for ethics
Get PDFHumans are reportedly exploited as the main attack vector for security breaches. In order to minimize the susceptibility of humans to security at-tacks, it is not sufficient for individuals to just be aware, but they need to change their behavior as well. Such behavior change, that is, the modification of user behavior, can occur via targeted interventions, which are gradually being introduced in cyber security. In this paper, we identify and categorize the main approaches used to change user behavior and portray the main limitations of these approaches. Other fields, like health sciences, psychology and economics, have been traditionally more mature in ethics-related considerations. We suggest that although individual behavior change is increasingly being embraced by security practitioners and professionals, ethical aspects of the accompanied interventions are by large neglected in the field. We explore the ethical traditions of utilitarian, deontological and virtue ethics and their relations with security. We posit that ethical frameworks are needed for cyber behavior change interventions as a means to enhance security hygiene on both an individual and an organizational level
Neural Inference Search for Multiloss Segmentation Models
Get PDFSemantic segmentation is vital for many emerging surveillance applications, but current models cannot be relied upon to meet the required tolerance, particularly in complex tasks that involve multiple classes and varied environments. To improve performance, we propose a novel algorithm, Neural Inference Search (NIS), for hyperparameter optimisation pertaining to established deep learning segmentation models in conjunction with a new multi-loss function. It incorporates three novel search behaviours, i.e. Maximised Standard Deviation Velocity Prediction, Local Best Velocity Prediction, and n-dimensional Whirlpool Search. The first two behaviours are exploratory, leveraging Long Short-Term Memory (LSTM)-(Convolutional Neural Network) CNN based velocity predictions, while the third employs n-dimensional matrix rotation for local exploitation. A scheduling mechanism is also introduced in NIS to manage the contributions of these three novel search behaviours in stages. NIS optimises learning and multi-loss parameters simultaneously. Compared with state-of-the-art segmentation methods and those optimised with other well-known search algorithms, NIS-optimised models show significant improvements across multiple performance metrics on five segmentation datasets. NIS also reliably yields better solutions as compared with a variety of search methods for solving numerical benchmark functions
Investigating epidemiological questions using Crithidia bombi and Bombus terrestris as a model system
Get PDFMedical Image Classification Using Transfer Learning and Network Pruning Algorithms
Get PDFDeep neural networks show great advancement in recent decades in classifying medical images (such as CTscans) with high precision to aid disease diagnosis. However, the training of deep neural networks requires significant sample sizes for learning enriched discriminative spatial features. Building a high quality dataset large enough to satisfy model training requirement is a challenging task due to limited disease sample cases, and various data privacy constraints. Therefore in this research, we perform medical image classification using transfer learning based on several well-known deep networks, i.e. GoogLeNet, Resnet and EfficientNet. To tackle data sparsity issues, a Wasserstein Generative Adversarial Network (WGAN) is used to generate new medical image samples to increase the numbers of training instances of the minority classes. The transfer learning process itself also allows the building of strong classifiers by transferring knowledge from the pre-trained image domain to a new medical domain using a small sample size. Moreover, the lottery ticket hypothesis is also used to prune each transfer learning network trained using the new target image data sets. Specifically, the L1 norm unstructured pruning technique is used for network reduction. Hyper-parameter finetuning is also performed to identify optimal settings of key network hyper-parameters such as learning rate, batch size and weight decay. A total of 20 trials are used for optimal hyper-parameter selection. Evaluated using multi-class lung X-ray images for pneumonia conditions and brain tumor CT-scans, the fine-tuned EfficientNet model obtains the best brain tumor classification accuracy rate of 96% and a fine-tuned GoogLeNet model with pruning has the highest pneumonia classification accuracy rate of 81.5%.<br/
Data sources:
Access Repository Dashboard
Do you manage Royal Holloway - Pure? Access insider analytics, issue reports and manage access to outputs from your repository in the CORE Repository Dashboard!