Deep Learning
Neural networks from neurons to architectures
Build and train neural networks in your browser. Backpropagation, activations, CNNs, and why depth matters.
Intermediate4 modules · 11 lessons~99 min950 XPAligned with fast.ai Practical Deep LearningStart with: ML FoundationsStart with: Math for AI
Progress0/11
Module
Neural Networks
From one neuron to deep stacks
Neurons & Layers9 min · 80 XPNeurons & Layers
Backpropagation10 min · 90 XPBackpropagation
Module
Training Deep Networks
The knobs and tricks that make depth actually trainable
Activation Functions9 min · 85 XPActivation Functions
Learning Rate & Schedules9 min · 85 XPLearning Rate & Schedules
Overfitting Deep Nets10 min · 90 XPOverfitting Deep Nets
Module
Computer Vision
Convolutions, the ImageNet moment, and the fine-tuning playbook
How CNNs See10 min · 95 XPHow CNNs See
Pooling & Architecture8 min · 80 XPPooling & Architecture
Transfer Learning10 min · 95 XPTransfer Learning
Module
Sequences & Beyond
Order, memory, and the hardware that made deep learning win
Why Vanilla Nets Fail on Sequences8 min · 85 XPWhy Vanilla Nets Fail on Sequences
Embeddings for Words8 min · 85 XPEmbeddings for Words
The Hardware Lottery8 min · 80 XPThe Hardware Lottery
Capstone project
Build a net by hand
Train a small neural network on MNIST or a similar small dataset. Ablation is the point: what happens with one layer? Bad activation? No normalization? Show it.
Deliverable: A notebook with training curves for at least three architecture variations and your analysis.
Milestones
Finish every lesson to earn the Deep Learning Certificate.