Course Wiki

Neural Networks

The deep learning foundations module: layers, activations, and training an image classifier with PyTorch, TensorFlow, and Keras.

Module 8 builds the deep learning picture from the inside: a neuron is a linear model passed through a nonlinearity, layers stack neurons, and backpropagation adjusts the weights. The course then moves to the frameworks — implementing the same model in TensorFlow/Keras and PyTorch to show what each abstraction does.

The applied project is image classification: convolutional layers for images, and the practical tricks that make training work — reading batches, checkpointing the best model, adding dropout, and augmenting training images to reduce overfitting.

Taught in

• Machine Learning Zoomcamp

  • Module 1: Introduction to Machine Learning
    • Model Selection Process
  • Module 8: Neural Networks and Deep Learning
    • Fashion classification
    • TensorFlow and Keras
    • Pre-trained convolutional neural networks
    • Convolutional neural networks
    • Transfer learning
    • Regularization and dropout
  • Module 9: Serverless Deep Learning
    • Introduction to Serverless
    • Python 3.12 vs TF Lite 2.17

• LLM Zoomcamp

  • Module 1: Agentic RAG
    • Introduction
  • Module 2: Vector Search
    • Vector Search ## Related concepts

• Transfer Learning

• Serverless Deployment

• Gradient Boosting

• Classification Metrics