International Journal of Artificial Intelligence

Enhancing Hydropower Management through Artificial Intelligence: Insights from Norway's Experience

Claude Jeroen — 2024-06-25

Norway is a global leader in renewable energy, with hydropower accounting for 90% of its electricity generation. The country's hydropower sector is crucial to both national and international energy demands, and the need for efficient management has become more pressing as the world shifts from fossil fuels to cleaner energy sources. Artificial Intelligence (AI) is emerging as a powerful tool for optimizing hydropower management by improving predictive analytics, automating decision-making, and processing real-time data. In Norway, AI is increasingly being used to forecast water flow and manage energy production more effectively, while also enhancing predictive maintenance to minimize downtime and operational costs. Despite its potential, the implementation of AI faces challenges such as high costs, infrastructure investments, and data privacy concerns. This article explores recent innovations in AI applied to hydropower in Norway, discussing both the opportunities and challenges. The successful integration of AI into hydropower operations holds promise for improving efficiency and sustainability, offering insights for broader adoption across the global renewable energy sector. Future developments in AI and its application in renewable energy, such as smart grids and interconnecting different energy sources, could further enhance the energy landscape.

EEG-based Classifications of Alzheimer’s Disease by Using Machine Learning Techniques

Nagarathna C. R — 2024-06-25

The study has shown how classifiers behave when identifying and categorizing Alzheimer's disease stages. The main characteristics of various frequency bands were fed into the classifier as input. The accuracy of recognition is evaluated using machine learning classifiers. The effort aims to create a novel model that combines "preprocessing, feature extraction, and classification" to identify different stages of disease. The study starts with bands filtering, moves on to feature extraction, which derives several bands from the EEG signals, and then employs KNN, SVM, and MLP algorithms to measure classification performance. "AD detection and classification using machine learning classifiers KNN, SVM, and MLP" is the main focus of this research. Five wavelet band characteristics are used by the built classifiers to categorize different illness phases. These characteristics are computed using DWT, PCA, and ICA, which aid in obtaining wavelet-related knowledge for learning. The proposed machine learning model achieves a classification accuracy of 95% overall.

Drone Based Fire Detection System Based on Convolutional Neural Network

Hanif Ikmal Rahman — 2024-06-25

Open fires are happening more and more throughout Malaysia. It is either intentional or accidental fire. The most dangerous is an accidental fire because it may not be detected by anyone until it becomes large. Detecting a fire is not an easy task. People may not see an ongoing fire because it may be too far away, or the fire may be too small. The objective of this project is to build a fire detection system. Fire detecting systems are developed to ensure more accurate fire detection. To ensure accurate fire detection, this project uses a waterfall methodology. This project uses drones as a tool to help with fire detection. Using a Convolutional Neural Network (CNN), this project implements the use of the PyTorch framework in detecting fires. The testing was done with a distance of 2 meters from the fire and a height of 2 meter from the ground. Edited images were used and uploaded to the system. Accuracy results of 80% can ensure accurate fire detection. To evaluate the system, edited fire images are used to ensure the accuracy of the system. Therefore, CNN is a good tool for detecting fires.

The Role of Artificial Intelligence in Disaster Prediction, Mitigation, and Response in the Philippines: Challenges and Opportunities

Rommel Baltazar — 2024-06-25

One of the most disaster-prone countries globally, experiences frequent natural calamities, including typhoons, earthquakes, and floods is the Philippines. This study explores the role of AI in enhancing disaster prediction, risk management, and mitigation in the Philippines. Using a qualitative research approach, semi-structured interviews were conducted between June 2023 and March 2024 with key stakeholders, including disaster management officials, meteorologists, and researchers. The findings highlight how AI technologies, particularly machine learning and neural networks, have significantly improved disaster forecasts by processing extensive datasets from meteorological, seismic, and geographical sources. AI-driven models are enhancing the accuracy of predictions for typhoons, earthquakes, and flood risks, contributing to more effective early warning systems and timely evacuation protocols. Despite these advancements, challenges remain, including limitations in infrastructure, budget constraints, and data quality, which hinder the full adoption of AI in disaster risk management (DRM). Nevertheless, the study identifies substantial opportunities for further development, emphasizing international collaboration and policy support to promote AI integration in DRM. The findings suggest that AI holds immense potential to revolutionize disaster response strategies in the Philippines, and further research is needed to address technical barriers and enhance AI’s role in building resilient communities.

An Innovative Technique for Medical Image Segmentation Using Convolutional Neural Networks Optimized Through Stochastic Gradient Descent

Mohammad Taheri — 2024-06-25

Medical image segmentation is crucial due to its essential role in disease therapy. Various challenges such as hair artifacts, illumination variations, and different imaging acquisitions complicate this task. In this paper, we introduce a novel convolutional neural network (CNN) architecture designed to address these challenges. We also compare our method with two well-known architectures, Unet and FCN, to demonstrate the superiority of our approach. Our results, evaluated using four metrics, accuracy, dice coefficient, Jaccard index, and sensitivity show that our method outperforms the other two. We employed Jaccard distance and binary cross-entropy as the loss functions and used SGD+Nesterov as the optimization algorithm, which proved more effective than the Adam optimizer. In the preprocessing step, we included image resizing to speed up the process and image augmentation to enhance the evaluation metrics. As a postprocessing step, we applied a threshold technique to the algorithm's outputs to increase the contrast of the final images. This method was tested on two well-known and publicly available medical image datasets: PH2 for melanoma detection and Chest X-ray images for detecting chest lesions or COVID-19.