Convolution Neural Network- an Advanced Deep Learning Algorithm for Image Classification and Performance Evaluation
Akshatha S.A
Computer Science and Engineering,
Bearys Institute of Technology, Mangalore
Email address: 69aksa@gmail.com (Akshatha S.A)
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - Artificial Intelligence is frequently described as an intelligent system equipped with the capacity for logical thinking and problem-solving, akin to a human being, but without genuine consciousness. The origin of AI dates back 8,000 to 11,000 years, with stories of artificial beings endowed with consciousness by ancient rishis of India. Machine Learning is often defined as "the capacity to acquire knowledge without the need for explicit programming." Deep Learning is part of ML consisting of algorithms that allow software to train itself to do the complex tasks such as speech recognition and image recognition, among others. An artificial neural network is an interconnected group of nodes that mimics the function of neurons in the brain. Deep learning employs artificial neural networks to conduct intricate computations on extensive datasets. Deep learning supports Convolutional Neural Networks (CNNs) to recognize and classify images. CNNs are a form of supervised learning, reliant on labeled data for training neural networks to identify and categorize images. In the present study, a Convolutional Neural Network algorithm was employed on a specific dataset, and its performance was assessed. The implementation followed a sequential approach within a Jupyter Notebook. The objective of the Convolutional Neural Network models was to discern between cats and dogs, involving two distinct classes. Ultimately, the model achieved a dataset accuracy of 81.25%.
Key Words: Convolution Neural Network, Artificial Intelligence, sequential model, Convolution layer, Image detection, dataset.
