This paper describes an unobtrusive sleep-wake monitor based on the wireless Respeck sensor worn as a patch on the chest, coupled with machinelearning-based classification methods for the automatic classification of ...
详细信息
ISBN:
(纸本)9798350387452;9798350387445
This paper describes an unobtrusive sleep-wake monitor based on the wireless Respeck sensor worn as a patch on the chest, coupled with machinelearning-based classification methods for the automatic classification of Sleep-Wake states in the Respeck sensor data. The \Sleep-Wake monitor was evaluated on labelled data collected by a cohort of 20 healthy volunteers and achieved results for accuracy, sensitivity and specificity of greater than 90% using Linear SVM, and nonlinear SVM-RBF models. Metrics such as frequency of Wake-After-Sleep-Onset (WASO) episodes were calculated using Respeck data to mark the quality of sleep in the cohort.
machinelearning (ML) has seen widespread adoption across different domains and is used to make critical decisions. However, with profuse and diverse data available, collaboration is indispensable for ML. The traditio...
详细信息
ISBN:
(纸本)9798350381993;9798350382006
machinelearning (ML) has seen widespread adoption across different domains and is used to make critical decisions. However, with profuse and diverse data available, collaboration is indispensable for ML. The traditional centralized ML for collaboration is susceptible to data theft and inference attacks. Federated learning (FL) promises secure collaborative machinelearning by moving the model to the data. However, FL faces the challenge of data and model poisoning attacks. This is because FL provides autonomy to the participants. Many Byzantine-robust aggregation schemes exist to identify such poisoned model updates from participants. But, these schemes require raw access to the local model updates, which exposes them to inference attacks. Thus, the existing FL is still insecure to be adopted. This paper proposes the very first generic FL framework, which is both resistant to inference attacks and robust to poisoning attacks. The proposed framework uses hyperdimensional computing (HDC) coupled with FL, called HDFL. HDFL is compatible with different (ML) model architectures and existing Byzantine-robust defenses. HDFL restricts drop in accuracy to 1-2%. HDFL does not add any additional communication overheads and incurs negligible computational time in encoding and decoding raw local model updates. Empirical evaluation demonstrates the effectiveness of HDFL. HDFL performs secure aggregation and achieves no-attack accuracy, even in the presence of 40% attackers, in just 1.2s per iteration.
An efficient oil palm plantation management involves monitoring the health of the trees to ensure good yield. However, manual inspection is unable to meet the expectation of large scale plantations. The feasibility of...
详细信息
ISBN:
(纸本)9798350349955;9798350349948
An efficient oil palm plantation management involves monitoring the health of the trees to ensure good yield. However, manual inspection is unable to meet the expectation of large scale plantations. The feasibility of employing unmanned aerial vehicles (UAV) and various artificial intelligence techniques have been studied with encouraging results. However, identifying unhealthy trees in UAV images is challenging. This research investigated the feasibility of using machinelearning techniques in distinguishing unhealthy trees from healthy trees. Four classifiers were evaluated with two sets of data, in which one is publicly available and the other was provided by our industry collaborator. Support vector machine (SVM) with different kernels was found to perform the best in both datasets, with an accuracy of 0.96 and 0.975 respectively.
Binary code similarity detection (BCSD) has numerous applications, including malware detection, vulnerability search, plagiarism detection, and patch identification. Recent studies have demonstrated that with the rapi...
详细信息
ISBN:
(纸本)9798350381993;9798350382006
Binary code similarity detection (BCSD) has numerous applications, including malware detection, vulnerability search, plagiarism detection, and patch identification. Recent studies have demonstrated that with the rapid progress of machinelearning (ML) techniques, various BCSD approaches based on machinelearning have exhibited stronger performance than traditional methods. However, current ML-based BCSD approaches tend to ignore the issue of training samples, and most ML-based BCSD approaches are based on supervised learning, which is suffered from the labelling difficulties. To mitigate these issues, we propose ConFunc: a function-level binary code similarity detection framework based on contrastive learning. Performance evaluation shows that ConFunc enhances the Mean Reciprocal Rank (MRR) and Recall rates (Recall@1) of baseline models by fully harnessing the potential of the data. Additionally, ConFunc demonstrates stronger performance in scenarios with scarce data, achieving the baseline model's performance on the entire dataset using only 10% of the complete dataset. In real-world patch identification and vulnerability search tasks, ConFunc consistently outperforms other baseline models in MRR and Recall@10.
Ensuring privacy while outsourcing the training of machinelearning (ML) models to cloud-based platforms is a critical concern. Although cryptographic solutions have been proposed, they often result in a substantial r...
详细信息
ISBN:
(纸本)9798350381993;9798350382006
Ensuring privacy while outsourcing the training of machinelearning (ML) models to cloud-based platforms is a critical concern. Although cryptographic solutions have been proposed, they often result in a substantial reduction in training accuracy and require modifications to the backend architecture. In this paper, we address the challenge of developing privacy-preserving techniques that offer adequate privacy without significantly impacting the accuracy of the ML model or the accuracy of the training process. We demonstrate that training private datasets on existing cloud-based platforms can be achieved with a high level of privacy and at a minimal cost in accuracy.
The increasing popularity of unsupervised machinelearning techniques, particularly in clustering algorithms, is evident due to their ability to efficiently generate clusters from large datasets. As data volumes conti...
详细信息
ISBN:
(纸本)9798350372977;9798350372984
The increasing popularity of unsupervised machinelearning techniques, particularly in clustering algorithms, is evident due to their ability to efficiently generate clusters from large datasets. As data volumes continue to expand, traditional methods become less feasible, prompting the exploration of parallel computing solutions for enhanced performance. This paper assesses the efficacy of parallel computing, focusing on Fuzzy C-Means clustering. Three implementations are compared: Sequential, Parallel using MPI, and Parallel using the Cloud. The adoption of parallel computing significantly improves scalability, leading to a 50% reduction in processing time and a 30% enhancement in overall system performance.
The eight papers included in this special issue represent a selection of extended contributions presented at the 17th internationalconference on Soft computing Models in Industrial and Environmental Applications, SOC...
详细信息
The eight papers included in this special issue represent a selection of extended contributions presented at the 17th internationalconference on Soft computing Models in Industrial and Environmental Applications, SOCO 2022 held in Salamanca, Spain, September 6th-8th, 2022, and organized by the BISITE group at University of Salamanca. SOCO 2022 internationalconference represents a collection or set of computational techniques in machinelearning, computer science and some engineering disciplines which investigate, simulate, and analyse very complex issues and phenomena. This special issue is aimed at practitioners, researchers, and postgraduate students who are engaged in developing and applying advanced intelligent systems principles to solve real-world problems in the mentioned fields.
ThirstyAI and resource-constrained environments have become evergrowing challenges for machinelearning (ML), especially large-scale Deep learning (DL) models. This paper introduces NanoDeploy Automator, an Automated ...
详细信息
ISBN:
(纸本)9798350385328;9798350385335
ThirstyAI and resource-constrained environments have become evergrowing challenges for machinelearning (ML), especially large-scale Deep learning (DL) models. This paper introduces NanoDeploy Automator, an Automated machinelearning (AutoML) subsystem for the TANDEM native Artificial Intelligence (AI) edge platform, designed for efficient and reliable Tiny machinelearning (TinyML), addressing the need for smaller models with minimal loss of accuracy and automated model compression and deployment at the edge. This work has been conducted in scope of project TANDEM and is currently expanded and adapted, forming the basis for the TinyML mechanisms of project SUNRISE-6G. The paper showcases the novel application of Time Generative Pre-Trained Transformer (TimeGPT) in real-world Internet of Things (IoT) scenarios, emphasizing three key contributions: a cutting-edge Automated TinyML (AutoTinyML) pipeline, the integration of TimeGPT in IoT, and a comprehensive evaluation of time series forecasting and anomaly detection techniques. Evaluation results demonstrate the efficiency of the AutoTinyML pipeline in terms of accuracy and model compression. The proposed system holds promise for diverse applications, leveraging the power of AutoTinyML, and reflects a significant stride towards democratizing access to accurate predictions and reducing uncertainty in edge computing scenarios, eliminating the need for a dedicated team of ML engineers.
This work provides a unique technique that uses advanced machinelearning (ML) algorithms with information collected from Fiber Bragg Grating (FBG) sensors to predict the temperature-strain connection. For strain sens...
详细信息
ISBN:
(纸本)9798350385939;9798350385922
This work provides a unique technique that uses advanced machinelearning (ML) algorithms with information collected from Fiber Bragg Grating (FBG) sensors to predict the temperature-strain connection. For strain sensing and structural health monitoring applications to be precise, FBG sensors are necessary. Developing a prediction model that can accurately and effectively anticipate changes in temperature-induced strain and predict a certain strain value for a given temperature is the aim of this study's integration of machinelearning methods. This method may be used to increase the effectiveness of structural health monitoring while also deploying sensors more cheaply and effectively. Fiber Bragg grating sensors are employed in the experimental stage to meticulously collect real-time data, including temperature and strain measurements. In order to optimize the training process of machinelearning models, a multifaceted method is employed to harvest and preprocess pertinent data. The prediction model's accuracy is ascertained utilizing a range of machinelearning techniques, including support vector machines, polynomial regression, decision tree models, and linear regression. The models are extensively trained and assessed, with a focus on assessing how effectively the models forecast temperature-induced fluctuations in strain. Among the models that were looked at, the Decision Tree model sticks out because its mean square error was the lowest when compared to the other models.
The complexity of financial systems and the rapid increase in data volume are making the use of intelligent computing and trustworthy machinelearning more important in finance. This paper discusses how intelligent co...
详细信息
ISBN:
(纸本)9798350377859;9798350377842
The complexity of financial systems and the rapid increase in data volume are making the use of intelligent computing and trustworthy machinelearning more important in finance. This paper discusses how intelligent computing can be applied within complex financial systems and takes a deeper look at the theory behind trustworthy machinelearning and how it is used in finance. By combining the structure of complex networks with the computing power of machinelearning, the paper also explores the inner workings of large neural networks and considers how to apply the theory of dynamic systems to the tuning of these networks, to improve intelligent computing in complex financial systems. Experiments based on the "scientific intelligence + machine conjecture" approach were carried out for risk assessment and market forecasting. The results show that these technologies can improve how financial institutions manage risk, help investors get more reliable information about market trends, and meet the transparency needed for regulatory compliance. The use of intelligent computing and trustworthy machinelearning in complex financial systems points to a future with lots of potential for innovation and new opportunities.
暂无评论