Start here
IOE414 Deep Learning endsem notes, from perceptrons to GANs.
How to use this siteSearch on YouTube
- Learn in order. Each lecture uses the one before it. The order is: foundations, training, CNNs, sequences, generative models.
- Work every example by hand before you open the solution. The endsem numericals match these types (PLA, GD steps, backprop, convolution, Bayes nets, VAE, GAN).
- Do the self-test at the bottom of each page.
- The night before the exam, read only the Recap boxes and the formula cheatsheet.
Press Ctrl K to search every lecture. Each page has Copy Markdown and Open buttons to take the notes into ChatGPT or Claude.
The course mapSearch on YouTube
Foundations
Neurons, perceptrons, activation functions, and how gradient descent trains them. Lectures 1–3.
Training deep networks
Backpropagation by hand, regularization, and optimizers from momentum to Adam. Lectures 4–6.
Convolutional networks
Hand convolution, CNN building blocks, LeNet, AlexNet, and embeddings. Lectures 7–8.
Sequence models
RNNs, backprop through time, LSTM, GRU, and bidirectional models. Lectures 9–10.
Probabilistic & generative models
Bayes nets, Markov chains, RBMs, autoencoders, VAEs, and GANs. Lectures 11–14.
Revision
Formula cheatsheet, solved midsem 2025 and endsem 2025 papers. Last-hour revision.