Deep Learning Foundations and Tools

This unit develops core mathematical, architectural, and practical skills for building and experimenting with deep learning models. Students will learn foundational linear algebra operations, neural network building blocks (including convolutions and LSTMs), and the software environment and datasets needed to train and evaluate models for vision and sequence tasks.

Learning Objectives

• Analyze scalar, vector, and matrix operations and demonstrate how these computations implement forward and backward passes in simple neural networks
• Apply development tools and frameworks by installing and configuring Anaconda, TensorFlow/Keras, and example code to run baseline models on provided datasets
• Demonstrate understanding of convolutional neural networks and long short-term memory networks by implementing and training simple models for image classification and sequence memorization
• Evaluate the effects of dataset size, data augmentation, and model complexity on performance using tasks such as traffic-sign recognition, object perception, and pose approximation