This document provides a summary of a study on deep learning. It introduces artificial neural networks as the building blocks of deep learning architectures. Neural networks are modeled after the human brain and consist of interconnected nodes that learn patterns in data. Deep learning aims to develop human-level artificial intelligence. The document explains key concepts like activation functions, which introduce non-linearity, and backpropagation, which is used to train neural networks by minimizing error. It surveys popular deep learning models and their objectives, like convolutional neural networks for computer vision and recurrent neural networks for language.