Deep Convolutional – Generative Adversarial Network for Autism Prediction

Abstract

autism spectrum disorder (ASD) is a multifaceted developmental condition characterized by enduring challenges in social communication, restricted interests, and other associated symptoms. This presents a consistent behavioral pattern. It is important to note that autism is a lifelong condition, but the extent of functional impediment varies among individuals. Early signs of this condition can be identified by parents, caregivers, or pediatricians before a child reaches their first birthday. Extensive research underscores that early intervention correlates with improved outcomes for individuals with autism in their later lives. As a result, there is an increasing demand for ASD prediction in order to detect the condition at an early stage. The effort to detect autism spectrum disorder (ASD) using machine learning has reached a notable milestone, with a notable accuracy of roughly 86.5%. Machine learning's prowess in ASD prediction has been unwavering, giving increased accuracy with less data while gradually declining when presented with large datasets. To overcome such constraints and raise the accuracy criteria for ASD prediction, this study delves into the domain of deep learning, that focuses on learning and making predictions from vast and complicated datasets using artificial neural networks. Deep learning techniques learn representations or features directly from raw data, as opposed to typical machine learning algorithms, which rely on constructed features. In this ambitious project, we begin on a paradigm shift from traditional machine learning approaches to deep learning, leveraging the limitless potential of advanced artificial neural networks. The method to this prediction is to use the Deep Convolutional GAN deep learning algorithm, with the goal of achieving higher precision.