Course Description
One first step towards developing Artificial intelligence (AI)-powered machines is to efficiently and effectively process sensory data and abstract them with symbolic or numerical descriptions, thereby generalizing their concepts and benefiting higher-level perception and decision-making tasks. Deep learning, a subset of machine learning, has emerged as a powerful approach to achieving this goal, particularly in the domain of computer vision, and it has led to human-level performance in various visual recognition tasks.
This course is designed to provide a broad introduction of deep learning techniques and their application to computer vision problems. Students will explore cutting-edge deep neural networks and learn how to leverage them to analyze, interpret, and understand visual data. The topics includes:
- Linear classifiers
- Stochastic gradient descent
- Fully-connected networks
- Convolutional networks
- Recurrent networks
- Attention and transformers
- Object detection
- Image segmentation
- Generative models
- And others
