Convolutional Neural Networks in Action
summary
Builds a small convolutional network in Python using PyTorch and tracks runs with DVC on the MNIST dataset, showing concrete layer choices: conv layers with 1→8→16 channels, 3x3 kernels, ReLU and max-pooling, then linear heads. Runs experiments adjusting learning rate and channel counts (changing 8→9 or 16→10 triggers shape errors), reports a live run peaking near 83% accuracy, and emphasizes tuning hyperparameters and reproducible experiment logs.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:55Neural networks overview
- 3:23Basics of convolutional neural networks
- 6:36Max pooling and layers
- 8:21Types of CNNs and their applications
- 10:48CNNs vs neural networks
- 12:10CNN use cases
- 13:36Live demo setup and PyTorch
- 14:42Building CNN with PyTorch
- 17:52Forward pass and activations
- 19:17Training with DVC
- 22:31Key takeaways and closing remarks
