Paper Title
Auto Generative Adversarial Network using Crow Search Algorithm for Human Activity Recognition

Abstract
This paper presents a new technique of high-quality Human Activity Recognition (HAR) system using the Neural Architecture Search (NAS) as a combined technique using the Generative Adversarial Networks (GAN) and the Crow Search Algorithm (CSA). The new technique can improve the total accuracy of HAR system compared to any previously used methods such as Artificial Neural Network (ANN) and Convolutional Neural Network (CNN). The previously used methods could not classify between much of overlapped objects within some of data sets. The new technique used the wideness of the search space and the evolutionary search abilities to detect the overlapping objects in an accurate methodology. AutoMl tools are used broadly to improve the accuracy of the classical machine learning methods in detection and segmentation of the independent and overlapped objects. Keywords - Central Force Optimization (CFO); Modified Central Force Optimization (MCFO); Fuzzy Logic Controller (FLC); Side Lobe Level (SLL)