Diminishing share of synchronous generators in the power system demands that inverter-interfaced generation sources provide grid-forming capability and ancillary services conventionally offered by SGs. Recognizing the pivotal role of grid-forming (GFM) inverters in this context, the National Grid Electricity System Operator (NGESO) recently amended its grid code facilitating the establishment of an ancillary services market based on GFM plants. The Network frequency perturbation (NFP) plot is recommended in the amended grid code as a compliance testing tool for a GFM plant. The interpretation of the NFP plot in this grid code and associated NFP plot interpretation guide is based on the oversimplified small-signal model of GFM inverters, which only considers the low-frequency dynamics of the active power control (APC) loop. Although the resulting model and analysis is simple and intuitive with such considerations, it is not known how ignorance of the dynamic coupling between the APC and RPC loops, as well as the electromagnetic transient dynamics of GFM inverters, may affect the interpretation of the NFP plot. In order to address this challenge, the accurate small-signal model of a typical GFM inverter is established in this article, based on which the NFP plot is developed and compared with its simplified version proposed in NFP plot interpretation guide. The outcome of this work provides insights for practitioners on the frequency ranges and conditions under which the simplification of the NFP plot can or cannot be made for modelling and interpreting the NFP plot of a GFM inverter. Experimental results are provided to validate the theoretical analysis.

Chronic Kidney Disease (CKD) is one of the widespread Chronic diseases with no known ultimo cure and high morbidity. Research demonstrates that progressive Chronic Kidney Disease (CKD) is a heterogeneous disorder that significantly impacts kidney structure and functions, eventually leading to kidney failure. With the progression of time, chronic kidney disease has moved from a life-threatening disease affecting few people to a common disorder of varying severity. The goal of this research is to visualize dominating features, feature scores, and values exhibited for early prognosis and detection of CKD using ensemble learning and explainable AI. For that, an AIdriven predictive analytics approach is proposed to aid clinical practitioners in prescribing lifestyle modifications for individual patients to reduce the rate of progression of this disease. Our dataset is collected on body vitals from individuals with CKD and healthy subjects to develop our proposed AI-driven solution accurately. In this regard, blood and urine test results are provided, and ensemble tree-based machine-learning models are applied to predict unseen cases of CKD. Our research findings are validated after lengthy consultations with nephrologists. Our experiments and interpretation results are compared with existing explainable AI applications in various healthcare domains, including CKD. The comparison shows that our developed AI models, particularly the Random Forest model, have identified more features as significant contributors than XgBoost. Interpretability (I), which measures the ratio of important to masked features, indicates that our XgBoost model achieved a higher score, specifically a Fidelity of 98%, in this metric and naturally in the FII index compared to competing models.

The Transformer architecture of Generative AI is very successful in predicting the distribution of Riemann zeta zero counts on consecutive Gram intervals. We get accuracies of 0.998 in predicting a sequence of ten consecutive zero counts. We tested with two ranges of Riemann zeta zeros, t = 1012 and t = 1028. With special training for rare events, we can get essentially full prediction. This shows that applying the technique to more complex problems has great promise. We have used very minimal computer resources compared to typical models in language applications. With access to better resources, we can attack much more important problems.

Hallucinations in AI-generated content pose a significant challenge to the reliability and trustworthiness of advanced language models, particularly as they become increasingly integrated into decision-making processes across various domains. The novel concept of employing Monte Carlo simulations on token probabilities offers a robust framework for detecting hallucinations, enhancing the accuracy and reliability of AIgenerated content. The methodology involved generating multiple outputs for diverse prompts, calculating token probabilities, and performing Monte Carlo simulations to identify low-probability tokens indicative of hallucinations. Experimental results revealed varying frequencies of hallucinations across different domains, with everyday scenarios presenting the highest challenge. The probabilistic framework enabled a detailed analysis of the LLM's outputs, providing insights into its decision-making process and highlighting the efficacy of the Monte Carlo approach. The study's findings underscore the importance of refining LLMs to improve their performance and reliability in real-world applications, contributing to the ongoing efforts to mitigate hallucinations in AI-generated content.

Raven's Progressive Matrices (RPMs) have been widely used for measuring abstract reasoning and intelligence in humans. However for artificial learning systems, abstract reasoning remains a challenging problem. This paper investigates the potential of biologically inspired spiking neural networks in addressing this challenge. We focus on unsupervised learning since in most domains where human-competitive conceptual learning is important, unlabelled data is plentiful and the autonomous discovery and understanding of underlying patterns without explicit guidance is paramount. Our experiments show that the overall accuracy of our unsupervised networks significantly outperform some supervised methods. Our results also demonstrate that, unlike their non-spiking counterparts, spiking neural networks are able to extract and encode relational features without any explicit instruction, do not rely on labelled training data, and are able to simulate the human ability of navigating and deciphering complex data landscapes. Our results exhibit state-of-the-art performance for unsupervised learning on the RAVEN (67.0%) and I-RAVEN (40.3%) datasets, indicating that that spiking neural networks are well suited to unsupervised conceptual learning.

By quantizing the kinematic and gravitational time dilation using various step sizes, one obtains a set of weighted orbit paths. The precession associated with each weighted orbit path combines to provide the same answer as the classical analytical solution.

This paper introduces a dataset characterizing the parameters of a stepped piezoelectric cantilever beam using various geometric specifications. The dataset also shows the electric potential generated at each layer when exposed to a specific excitation frequency. The primary application of the cantilever is in ambient vibration energy harvesting, where the piezoelectric cantilever beam plays a crucial role in converting mechanical vibrations into electrical energy. By shedding light on the complex interaction between geometric features and electric potential distributions, this dataset is valuable for advancing knowledge in energy harvesting technologies. The availability of this dataset facilitates further analysis, model validation, and optimization efforts, contributing to the ongoing enhancement of ambient energy harvesting systems by optimizing geometries.

Non-electroencephalogram seizure detection models hold promise for the early detection of generalised onset seizures. However, these models often experience high false alarm rates and difficulties in distinguishing normal movements from seizure manifestations. To address this, we were granted exclusive access to the newly developed Open Seizure Database, from which a representative dataset of 94 events was selected (42 generalised tonic-clonic seizures, 19 auras/focal seizures, and 33 seizures labelled as Other), with a combined duration of approximately 5 hours and 29 minutes. Each event contains acceleration and heart rate data which was expertly annotated by a clinician, who labelled every 5 second timestep with a class of Normal, Pre-Ictal, or Ictal. We then introduced the AMBER (Attention-guided Multi-Branching-pipeline with Enhanced Residual fusion) model. AMBER constructs multiple branches to form independent feature extraction pipelines for each sensing modality. The outputs of each branch are passed to our custom Enhanced Residual Fusion layer, where the extracted features are combined into a fused representation. The fused representation is then propagated through two densely connected blocks before being passed through a softmax activation function. The model was trained using k-fold cross validation, with k-1 fold used to train the model and the remaining fold was used to evaluate the model’s performance. The results of these experiments underscore the efficacy of Ictal-Phase Detection, achieving an accuracy and f1-score of 0.8995 and 0.8987. Notably, the model exhibited consistent generalisation, recording a True Positive Rate of 0.9564, 0.8325, and 0.9111 for the Normal, Pre-Ictal, and Ictal classes respectively. These findings were compounded by an average False Positive Rate, recording an overall score of 0.0502. In conclusion, this research introduces a new detection technique and model designed for multimodal seizure detection, with the potential to reduce the false alarm window and differentiate high and low amplitude convulsive movement. We believe the results of this study lay the groundwork for further advancements in non-electroencephalogram seizure detection research.

The increasing sophistication and capabilities of artificial intelligence systems have brought about significant advancements in natural language processing, yet they have also exposed these systems to various security vulnerabilities, particularly targeted prompt injection attacks. The introduction of a moving target defence mechanism offers a novel and significant approach to mitigating these attacks through continuously altering the model’s parameters and configurations, thereby creating an unpredictable environment that complicates adversarial efforts. This research provides a comprehensive evaluation of the moving target defence mechanism, detailing the selection and categorization of prompt injection attacks, the development of dynamic defence techniques such as random parameter perturbation, model re-initialization, and dynamic context adjustments, and their seamless integration with the Mistral LLM. The experimental results indicate a substantial reduction in the attack success rate, maintaining high performance metrics while managing computational overhead efficiently. The findings highlight the practical applicability and potential for widespread adoption of the moving target defence mechanism in enhancing the security and resilience of large language models against sophisticated adversarial tactics.

Modern fifth-generation (5G) networks are increasingly moving towards architectures characterized by softwarization and virtualization. This paper addresses the complexities and challenges in deploying applications and services in the emerging multi-tiered 5G network architecture, particularly in the context of microservices-based applications. These applications, characterized by their structure as directed graphs of interdependent functions, are sensitive to the deployment tiers and resource allocation strategies, which can result in performance degradation and susceptibility to failures. Additionally, the threat of deploying potentially malicious applications exacerbates resource allocation inefficiencies. To address these issues, we propose a novel optimization framework that incorporates a probabilistic approach for assessing the risk of malicious applications, leading to a more resilient resource allocation strategy. Our framework dynamically optimizes both computational and networking resources across various tiers, aiming to enhance key performance metrics such as latency, accuracy, and resource utilization. Through detailed simulations, we demonstrate that our framework not only satisfies strict performance requirements but also surpasses existing methods in efficiency and security.

A document by Xiangdong Wei. Click on the document to view its contents.

This research investigated the effect of temperature on coconut oil extraction using a solvent extraction method. The study aimed to optimize the extraction process and provide insights into the relationship between temperature and oil yield. The research employed a laboratory-scale solvent extraction method, where coconut oil was extracted from shredded coconut at different temperatures (30°C, 40°C, 50°C, 60°C, and 70°C). The yield of oil extracted was measured and calculated as a percentage of the initial weight of coconut used. The results showed a significant increase in oil yield with increasing temperature, with the highest yield obtained at 70°C. The yield increased from 26.29% at 30°C to 32.70% at 70°C, indicating a 24.5% increase. The study demonstrates the importance of temperature optimization in coconut oil extraction, providing valuable insights for the industry. The findings suggest that increasing the temperature to 70°C can result in a higher yield and improved extraction efficiency. This research contributes to the existing body of knowledge on coconut oil extraction, highlighting the potential for process optimization and improved productivity. The results have important implications for coconut oil manufacturers, providing a basis for the development of more efficient and effective extraction processes. Overall, this study demonstrates the significance of temperature optimization in coconut oil extraction, providing a foundation for future research and industrial applications.

This study investigated the physicochemical properties and glycemic potential of two honey samples from Ikwuano (Umuahia) and Umuabiara Amii Akabo (Imo state). The research aimed to determine the sugar content, moisture content, pH, specific gravity, and glycemic load of the honey samples, providing valuable insights for consumers, particularly those with dietary restrictions or preferences. The results revealed significant differences in the physicochemical properties of the two honey samples. The moisture content of the Ikwuano honey was 29.5933%, while that of Umuabiara Amii Akabo was 30.4338%. The pH of Ikwuano honey was 4.28, indicating a slightly acidic nature, while that of Umuabiara Amii Akabo was 3.45, indicating a more acidic nature. The specific gravity of Ikwuano honey was 1.2680, while that of Umuabiara Amii Akabo was 1.2705, indicating a slight difference in density. The reducing sugar content was determined using lead acetate and potassium oxalate solutions, and the titration values for the Ikwuano honey and Umuabiara Amii Akabo honey were 22.0 and 19.0, respectively. The glucose content was also determined, and the glycemic load of the Ikwuano honey was found to be 46.12, while that of the Umuabiara Amii Akabo honey was 50.72.These findings suggest that the two honey samples have different glycemic potentials, with the Umuabiara Amii Akabo honey having a higher glycemic load. This information is crucial for individuals with dietary restrictions, such as those with diabetes, who need to monitor their sugar intake. Additionally, the study highlights the importance of analytical techniques in determining the quality and properties of honey samples.The research advances our understanding of the physicochemical properties and glycemic potential of honey samples from different regions, providing valuable insights for consumers, producers, and regulatory agencies. The study’s findings can inform the development of guidelines for honey production and labeling, ensuring that consumers have access to accurate information about the products they consume. Furthermore, the research contributes to the growing body of knowledge on the nutritional and health benefits of honey, highlighting its potential as a natural sweetener and functional food ingredient.

Language models are prone to generating hallucinations, which significantly undermine their reliability and usefulness in critical applications. Introducing a novel approach that combines semantic relevance scoring with K-means clustering, our methodology significantly enhances the model's accuracy and reduces the occurrence of hallucinations. By integrating these techniques, the model can prioritize contextually appropriate synonyms, resulting in more coherent and factually correct outputs. The experimental results demonstrate substantial improvements in accuracy, semantic relevance, and a marked reduction in hallucinations across various tasks. Comprehensive evaluation using diverse metrics demonstrates the robustness and effectiveness of our modifications, highlighting the potential for practical deployment in applications where accuracy and reliability are paramount. This study affirms the viability of combining semantic relevance scoring with clustering techniques to enhance the performance of language models, contributing to the development of more reliable and effective models for a wide range of applications.

A document by Junliang Yang. Click on the document to view its contents.

The Landau distribution is an important distribution in physical sciences that describes the statistical property of the energy loss by charged particles as they traversing a thin layer of matter. However, this distribution is difficult to treat analytically and requires integration for numerical evaluation. The CERNLIB-G110 approximation, which is currently widely used, is accurate to within 8-9 digits of the relative error. In this paper, an approximate formula with 16-digit relative error, which is accurate enough for double type of IEEE754, and an approximate formula with 34-digit relative error for float128 are obtained. In addition to the approximation, numerical evaluation of Landau distribution and random number generation method are also presented.

Mild steels composed by two main elements, they are iron (Fe) and carbon (C) elements which is widely used in industry because of its resistance and more affordable in terms of cost than stainless steel, but their weakness is that they have low corrosion resistance. One way to modify mild steel is by coating them with antioxidant compounds that can delay, slow down, and prevent lipid oxidation process, which is obtained from(Chrysophyllum-Albidum) Udara seed extract. This research work is aimed at producing organic corrosion inhibitor from(Chrysophyllum-Albidum) Udara seed extract. Also, to utilize this organic corrosion inhibitor as a substitute for chemical corrosion inhibitors which contain toxic compounds, and determining the corrosion inhibition efficiency of the Udara seed extract in Hydrochloric acid medium. This was carried out by weighing the mild steel pieces. Udara seed extract was also weighed and added into each of the transparent glass bottles that was used in the experiment. However, 10.0g Udara seed powder was added into bottle 1 containing 0.5m of the dilute tetraoxosulphate (vi) acid. 20.0g to bottle 2, containing the same concentration of the acid. 30.0g to bottle 3, 40.0g to bottle 4, containing 0.5m of the dilute tetraoxosulphate (vi) acid. And 50.0g to bottle 5 containing the same acid, and Finally, bottle 6 was containing the mild steel metal and the dilute tetraoxosulphate (vi) acid. It was observed that the Udara seed extract effectively inhibited mild steel corrosion in H2SO4 solution. Inhibition efficiency was observed to improve with increase in concentration of the extract. It was generally observed that inhibition efficiency was low at the first hour and with the blank and the first concentration and gradually increases by the preceding hours and concentrations. Weight loss increased with increase in time portraying retardation in the inhibitor efficiency with increase in time.

This paper proposes new methods to enhance clickthrough rate (CTR) prediction models using the Deep Interest Network (DIN) model, specifically applied to the advertising system of Alibaba's Taobao platform. Unlike traditional deep learning approaches, this research focuses on localized user behavior activation for tailored ad targeting by leveraging extensive user behavior data. Compared to traditional models, this method demonstrates superior ability to handle diverse and dynamic user data, thereby improving the efficiency of ad systems and increasing revenue.