Array最新文献

Semi-supervised clustering is a machine learning technique that was introduced to boost clustering performance when labeled data is available. Indeed, some labeled data are usually available in real use cases, and can be used to initialize the clustering process to guide it and to make it more efficient. Fuzzy ART is a clustering technique that is proved to be efficient in several real cases, but as an unsupervised algorithm, it cannot use available labeled data. This paper introduces a semi-supervised variant of the FuzzyART clustering algorithm (SSFuzzyART). The proposed solution uses the available labeled data to initialize clusters centers. In another hand, to deeply evaluate the characteristics of the proposed algorithm, a clustered binary data generation algorithm with controlled partitioning is also introduced in this paper. Indeed, the controlled generated clusters allows studying the characteristics of the proposed SSFuzzyART. Furthermore, a set of experiments is carried out on some available benchmarks. SSFuzzyART demonstrated better clustering prediction results than its classic counterpart.

The speed difference between high-performance CPUs and energy-efficient CPUs, which are found in asymmetric performance multicore processors, affects the current form of Amdahl’s law equation. This paper proposes two updates to that equation based on the performance evaluation results of a simple parallel pi program written with OpenCilk. Performance evaluation was done by measuring execution time and instructions per cycle (IPC). The performance evaluation of the parallel program executed on the Intel Core i5 1240P processor did not indicate decreased performance due to asymmetric performance. Instead, the program with efficient work-stealing advantages from OpenCilk performed well. In the case of using the execution time of the P-CPU as a reference to obtain speedup, the evaluation results in a sublinear speedup. Conversely, in the case of using the execution time of the E-CPU as a reference, the evaluation results in a superlinear speedup. This paper proposes two updates to Amdahl’s law equation based on these two evaluation results.

Electroencephalography (EEG) signals are used to evaluate the activities of the brain. For the accidents occurring on the road, one of the primary reasons is driver fatigueness and it can be easily identified by the EEG. In this work, five efficient and robust approaches for the EEG-based driving fatigue detection and classification are proposed. In the first proposed strategy, the concept of Multi-Dimensional Scaling (MDS) and Singular Value Decomposition (SVD) are merged and then the Fuzzy C Means based Support Vector Regression (FCM-SVR) classification module is utilized to get the output. In the second proposed strategy, the Marginal Fisher Analysis (MFA) is implemented and the concepts of conditional feature mapping and cross domain transfer learning are implemented and classified with machine learning classifiers. In the third proposed strategy, the concepts of Flexible Analytic Wavelet Transform (FAWT) and Tunable Q Wavelet Transform (TQWT) are implemented and merged and then it is classified with Extreme Learning Machine (ELM), Kernel ELM and Adaptive Neuro Fuzzy Inference System (ANFIS) classifiers. In the fourth proposed strategy, the concepts of Correntropy spectral density and Lyapunov exponent with Rosenstein algorithm is implemented and then the multi distance signal level difference is computed followed by the calculation of the Geodesic minimum distance to the Riemannian means and finally tangent space mapping is implemented to it before feeding it to classification. In the fifth or final proposed strategy, the Hilbert Huang Transform (HHT) is implemented and then the Hilbert marginal spectrum is computed. Then using the Blackhole optimization algorithm, the features are selected and finally it is classified with Cascade Adaboost classifier. The proposed techniques are applied on publicly available EEG datasets and the best result of 99.13% is obtained when the proposed Correntropy spectral density and Lyapunov exponent with Rosenstein algorithm is implemented with the multi distance signal level difference followed by the calculation of the Geodesic minimum distance to the Riemannian means and finally tangent space mapping is implemented with Support Vector Machine (SVM) classifier.

In this article a real-time method is proposed that reduces the pose estimate error for robots capable of motion on the 2D plane. The solution that the method provides addresses the recent introduction of low-cost panoramic range scanners (2D LIDAR range sensors whose field of view is 360${}^{\circ }$), whose use in robot localisation induces elevated pose uncertainty due to their significantly increased measurement noise compared to prior, costlier sensors. The solution employs scan-to-map-scan matching and, in contrast to prior art, its novelty lies in that matching is performed without establishing correspondences between the two input scans; rather, the matching problem is solved in closed form by virtue of exploiting the periodicity of the input signals. The correspondence-free nature of the solution allows for dispensing with the calculation of correspondences between the input range scans, which (a) becomes non-trivial and more error-prone with increasing input noise, and (b) involves the setting of parameters whose output effects are sensitive to the parameters’ correct configuration, and which does not hold universal or predictive validity. The efficacy of the proposed method is illustrated through extensive experiments on public domain data and over various measurement noise levels exhibited by the aforementioned class of sensors. Through these experiments we show that the proposed method exhibits (a) lower pose errors compared to state of the art methods, and (b) more robust pose error reduction rates compared to those which are capable of real-time execution. The source code of its implementation is available for download.

Healthcare is experiencing a fast change from a hospital-centric and specialist-focused model to one that is dispersed and patient-centric. Numerous technological advancements are driving this fast evolution of the healthcare sector. Communication technologies, among others, have permitted the delivery of customized and distant healthcare services. The present 4G networks and other wireless communication technologies are being utilized by the healthcare industry to create smart healthcare applications. These technologies are continuously evolving to meet the expectations and requirements of future smart healthcare applications. At the moment, current communication technologies are incapable of meeting the dynamic and complex demands of smart healthcare applications. Thus, the future 5G and beyond 5G networks are expected to support smart healthcare applications such as remote surgery, tactile internet and Brain-computer Interfaces. Future smart healthcare networks will combine IoT and advanced wireless communication technologies that will address current limitations related to coverage, network performance and security issues. This paper presents 5G-based smart healthcare architecture, key enabling technologies and a deep examination of the threats and solutions for maintaining the security and privacy of 5G-based smart healthcare networks.

There are various types of obstacles in an emergency, and the traffic environment is complicated. It is critical to detect obstacles accurately and quickly in order to improve traffic safety. The obstacle detection algorithm based on deep learning cannot detect all types of obstacles because it requires pre-training. The VIDAR (Vision-IMU-based Detection and Range method) can detect any three-dimensional obstacles, but at a slow rate. In this paper, an improved VIDAR and machine learning-based obstacle detection method (hereinafter referred to as the IVM) is proposed. In the proposed method, morphological closing operation and normalized cross-correlation are used to improve VIDAR. Then, the improved VIDAR is used to quickly match and remove the detected unknown types of obstacles in the image, and the machine learning algorithm is used to detect specific types of obstacles to increase the speed of detection with the average detection time of 0.316s. Finally, the VIDAR is used to detect regions belonging to unknown types of obstacles in the remaining regions, improving detection performance with the accuracy of 92.7%. The flow of the proposed method is illustrated by the indoor simulation test. Moreover, the results of outdoor real-world vehicle tests demonstrate that the method proposed in this paper can quickly detect obstacles in real-world environments and improve detection accuracy.

Depression is a common psychiatric disorder that is becoming more prevalent in developing countries like Bangladesh. Depression has been found to be prevalent among youths and influences a person’s lifestyle and thought process. Unfortunately, due to the public and social stigma attached to this disease, the mental health issue of individuals are often overlooked. Early diagnosis of patients who may have depression often helps to provide effective treatment. This research aims to develop mechanisms to detect and predict depression levels and was applied to university students in Bangladesh. In this work, a questionnaire containing 106 questions has been constructed. The questions in the questionnaire are primarily of two kinds – (i) personal, and (ii) clinical. The questionnaire was distributed amongst Bangladeshi students and a total of 684 responses (aged between 19 and 35) were obtained. After appropriate consents from the participants, they were allowed to take the survey. After carefully scrutinizing the responses, 520 samples were taken into final consideration. A hybrid depression assessment scale was developed using a voting algorithm that employs eight well-known existing scales to assess the depression level of an individual. This hybrid scale was then applied to the collected samples that comprise personal information and questions from various familiar depression measuring scales. In addition, ten machine learning and two deep learning models were applied to predict the three classes of depression (normal, moderate and extreme). Five hyperparameter optimizers and nine feature selection methods were employed to improve the predictability. Accuracies of 98.08%, 94.23%, and 92.31% were obtained using Random Forest, Gradient Boosting, and CNN models, respectively. Random Forest accomplished the lowest false negatives and highest F Measure with its optimized hyperparameters. Finally, LIME, an explainable AI framework, was applied to interpret and retrace the prediction output of the machine learning models.

Breast cancer, lung cancer, skin cancer, and blood malignancies such as leukemia and lymphoma are just a few instances of cancer, which is a collection of cells that proliferate uncontrollably within the body. Acute lymphoblastic leukemia is of one the significant form of malignancy. The hematologists frequently makes an oversight while determining a blood cancer diagnosis, which requires an excessive amount of time. Thus, this research reflects on a novel method for the grouping of the leukemia with the aid of the modern technologies like Machine Learning and Deep Learning. The proposed research pipeline is occupied into some interconnected parts like dataset building, feature extraction with pre-trained Convolutional Neural Network (CNN) architectures from each individual images of blood cells, and classification with the conventional classifiers. The dataset for this study is divided into two identical categories, Benign and Malignant, and then reshaped into four significant classes, each with three subtypes of malignant, namely, Benign, Early Pre-B, Pre-B, and Pro-B. The research first extracts the features from the individual images with CNN models and then transfers the extracted features to the features selections such as Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and SVC Feature Selectors along with two nature inspired algorithms like Particle Swarm Optimization (PSO) and Cat Swarm Optimization (CSO). After that, research has applied the seven Machine Learning classifiers to accomplish the multi-class malignant classification. To assess the efficacy of the proposed architecture a set of experimental data have been enumerated and interpreted accordingly. The study discovered a maximum accuracy of 98.43% when solely using pre-trained CNN and classifiers. Nevertheless, after incorporating PSO and CSO, the proposed model achieved the highest accuracy of 99.84% by integrating the ResNet50 CNN architecture, SVC feature selector, and LR classifiers. Although the model has a higher accuracy rate, it does have some drawbacks. However, the proposed model may also be helpful for real-world blood cancer classification.

With the proliferation of smartphones, digital data collection has become trivial. The ability to analyze images has increased, but source authentication has stagnated. Editing and tampering of images has become more common with advancements in signal processing technology. Recent developments have introduced the use of seam carving (insertion and deletion) techniques to disguise the identity of the camera, specifically in the child pornography market. In this article, we focus on the available features in the image based on PRNU (photo response nonuniformity). The forced-seam sculpting technique is a well-known method to create occlusion for camera attribution by injecting seams into each 50 × 50 pixel block. To counter this, we perform camera identification using a 1D CNN integrated with feature extractions on 20 × 20 pixel blocks. We achieve state-of-the-art performance for our proposed IMG_CAT (image categorization) in three-class classification over the baselines (original, seam removed, seam inserted). Based on our experimental findings, our model is robust when dealing with blind facts related to the questionable camera.

Recently, a novel type of Neural Network (NNs): the Physics-Informed Neural Networks (PINNs), was discovered to have many applications in computational physics. By integrating knowledge of physical laws and processes in Partial Differential Equations (PDEs), fast convergence and effective solutions are obtained. Since training modern Machine Learning (ML) systems is a computationally intensive endeavour, using Quantum Computing (QC) in the ML pipeline attracts increasing interest. Indeed, since several Quantum Machine Learning (QML) algorithms have already been implemented on present-day noisy intermediate-scale quantum devices, experts expect that ML on reliable, large-scale quantum computers will soon become a reality. However, after potential benefits from quantum speedup, QML may also entail reliability, trustworthiness, safety, and security risks. To solve the challenges of QML, we combine classical information processing, quantum manipulation, and processing with PINNs to accomplish a hybrid QML model named quantum based PINNs.