Deep Learning Hand Written Notes

Unleash Your Inner Genius: Master DL with Highly Curated Notes

I have carefully chosen and thoughtfully organised DL concepts from various online resources, books and a bunch of my Post-Grad courses at IISc Bangalore. Generally speaking, below are the concepts covered:

Foundations & feedforward networks

• Perceptron & perceptron learning algorithm
• Sigmoid neuron / activation functions
• Feedforward networks & function approximation
• Backpropagation (chain rule, computing gradients)

Optimization

• Gradient descent
• Momentum-based gradient descent
• Nesterov accelerated gradient (lookahead)
• RMSProp
• Adam
• Moving averages of gradients / learning rate

Linear algebra & PCA foundations

• Eigenvalue decomposition (EVD)
• Singular value decomposition (SVD)
• Principal component analysis (PCA)
• Low-rank / best rank-1 approximation

Autoencoders

• Undercomplete & overcomplete autoencoders
• Linear autoencoder ↔ PCA connection
• Regularized autoencoders
• Denoising & contractive autoencoders

Regularization & generalization

• Bias–variance tradeoff
• L2 regularization
• Adding noise (to inputs / outputs)
• Bagging / ensembles
• Dropout
• Weight initialization (Xavier / He)

Convolutional neural networks

• Convolution & pooling
• CNN architectures (e.g., Inception module)
• Visualizing & understanding CNNs (occlusion, feature visualization, generating images from embeddings)

Sequence models

• Recurrent neural networks (RNNs)
• Backpropagation through time (BPTT)
• Vanishing / exploding gradients
• LSTMs / GRUs (and why they mitigate vanishing gradients)
• Attention

Generative models

• Generative Adversarial Networks (GANs) — generator/discriminator objectives, minimax
• Variational Autoencoders (VAEs) — encoder/decoder, ELBO, inference

Don't hesitate and grab the deal. These notes helped me get offers from Adobe, Flipkart and Google !

If you need an exception or are not satisfied, feel free to reach out and request a refund (no questions asked).