Epileptic seizure is a neurological condition caused by short and unexpectedly occurring electrical disruptions in the brain. It is estimated that roughly 60 million individuals worldwide have had an epileptic seizure. Experiencing an epileptic seizure can have serious consequences for the patient. Automatic seizure detection on electroencephalogram (EEG) recordings is essential due to the irregular and unpredictable nature of seizures. By thoroughly analyzing EEG records, neurophysiologists can discover important information and patterns, and proper and timely treatments can be provided for the patients. This research presents a novel machine learning-based approach for detecting epileptic seizures in EEG signals. A public EEG dataset from the University of Bonn was used to validate the approach. Meaningful statistical features were extracted from the original data using discrete wavelet transform analysis, then the relevant features were selected using feature selection based on the binary particle swarm optimizer. This facilitated the reduction of 75% data dimensionality and 47% computational time, which eventually sped up the classification process. After having been selected, relevant features were used to train different machine learning models, then hyperparameter optimization was utilized to further enhance the models’ performance. The results achieved up to 98.4% accuracy and showed that the proposed method was very effective and practical in detecting seizure presence in EEG signals. In clinical applications, this method could help relieve the suffering of epilepsy patients and alleviate the workload of neurologists.

Edge Artificial Intelligence based traffic tracking and detecting sensors are very essential for smart cities, especially for smart transportation applications. These sensors are not only used to collect large amount of traffic data, but also reduce the bandwidth of the communication network to transfer and reduce the workload to process huge data on the cloud or server side. It is very necessary to process, store, and extract useful data at the edge of the Internet before transferring the data to central server which can be called Artificial Intelligence on The Edge. This research aims at studying, implementing and evaluating machine learning models which are suitable for running on limited computing embedded computers. Computer vision and real-time object detection techniques are applied on Nvidia Jetson Nano Embedded Computer to build an Edge-AI based traffic tracking and detecting sensor, two popular models (MobileNet-SSD and YOLOv4) have been studied and implemented to compare the performance on vehicle counting and license plate detection applications. There has been a new propose method to apply TensorRT engine for these two models to increase the processing speed. The data source used in this project is manually collected at actual traffic routes and parking lots in Vietnam with more than 11700 images of Vietnamese vehicles and license plates then trained on Google Colab. The performance evaluation results show that both models have high accuracy in real-time license plate detection and vehicle counting application when implemented on the Edge computer Jetson Nano with the mAPs of both model are higher than 90 percent during training session. The MobileNet-SSD model has good speed (40 FPS) which is very much faster than some previous works (25 FPS), this model is suitable for real-time applications. The YOLOv4 model, after being optimized by TensorRT engine, has a better speed (7.2 FPS) than the original version (1.7 FPS), although the YOLOv4 model has low speed than MobileNet-SSD model but can detect smaller size, this model is suitable for some applications that need to detect complicated objects with small sizes.

The main objective of this study is to develop and evaluate an effective integrated learning approach for the automatic classification of Municipal Solid Waste using the TrashBox data set comprising of 17,785 images to enhance the sorting of recyclable waste materials, curb on the use of landfills and enhance the attainment of sustainable environmental practices. Firstly, four deep learning models-DenseNet161, ResNet152, and MobileNetV3 variants are explored to identify the most suitable feature extraction method. In feature selection phase, three metaheuristic algorithms-Whale Optimization Algorithm (WOA), Salp Swarm Algorithm (SSA), and Harris Hawk Optimization (HHO)-are applied to remove the concept of less important features and keep significant features. The selected features are then fed to machine learning classifiers- Support Vector Machine (SVM), Random Forest (RF), Logistic Regression (LR) and K-Nearest Neighbor (KNN) algorithms- for final predictions. The variant DenseNet161-HHO-SVM performs better than other combinations in this study due to its highest accuracy and lowest execution time. This integrated approach also has better performance (97.45%) than previous state-of-the-art models on the same dataset while data processing and method integration phases make incredible impacts.

Dimensions, volume, and mass of agricultural goods are essential physical features to build sizing, grading, and packaging systems. Manual property measurements are time-consuming, costly, and labor-intensive due to traditional technologies while various devices need to be installed to adapt process requirements. Furthermore, it is very challenging to accurately measure product with irregular or specific shapes, such as starfruit (Averrhoa carambola). Recently, there have been several research results on estimating features of the irregularly shaped object, they are either get inaccurate results or need repeated captures, computational resources, and time to rebuild the three-dimensional representation of the goods. The starfruit has not been studied completely in this size and weight measurements. This paper focuses on new techniques which exhibit simple installation to generate multiple functions for estimating the dimensions, volume, and mass with high accuracy. In this proposed method, we separated the process into two main phases. In the first phase, a camera is used to capture a top-view image of a starfruit, then image processing and machine vision techniques are applied to process the acquisition image before the process slices numerically the starfruit into several pieces along the longitudinal axis and estimates the physical attributes of each pieces using disc method and conical frustum method. Its volume is the summation of the volume of each partial slice. In the next phase, the density is used to estimate the mass of starfruit since the correlation coefficient (R-squared) between the volume and mass of starfruit is nearly linear with 0.9205. The validated results are highly competitive with accuracy of about 99% for the volume and mass in 300 testing samples.

The development of new algorithms can aid researchers and professionals in resolving problems that were once unsolvable or discovering superior solutions to problems that were already settled. By recognizing the importance of continuous research on creating novel algorithms, this paper introduced a hybrid metaheuristic algorithm-h-PSOGNDO, which is a combination of Particle Swarm Optimization (PSO) and Generalized Normal Distribution Optimization (GNDO). The proposed algorithm utilizes the Particle Swarm Optimization’s strategy for exploitation and the Generalized Normal Distribution Optimization’s global search strategy for exploration. Through this combination, h-PSOGNDO is believed to be an effective algorithm that can promote the advantages of its parents’ algorithms. Different assessment methods are used to assess the proposed novel algorithm. First, the h-PSOGNDO is set to conduct experiments on two sets of mathematical functions, including twenty-eight IEEE CEC2017 and ten IEEE CEC2019 benchmark test functions, respectively. Then, the h-PSOGNDO algorithm is applied to a case study on the prediction of antimicrobial peptides’ toxicity to evaluate its performance on real-life problems. The statistical findings collected from both the test function sets and the case study show that the h-PSOGNDO algorithm works effectively, proving its astonishing ability to yield highly competitive outcomes for complex problems.