NOC:Deep Learning


Lecture 1 - Biological Neuron


Lecture 2 - From Spring to Winter of AI


Lecture 3 - The Deep Revival


Lecture 4 - From Cats to Convolutional Neural Networks


Lecture 5 - Faster, higher, stronger


Lecture 6 - The Curious Case of Sequences


Lecture 7 - Beating humans at their own games (literally)


Lecture 8 - The Madness (2013)


Lecture 9 - (Need for) Sanity


Lecture 10 - Motivation from Biological Neurons


Lecture 11 - McCulloch Pitts Neuron, Thresholding Logic


Lecture 12 - Perceptrons


Lecture 13 - Error and Error Surfaces


Lecture 14 - Perceptron Learning Algorithm


Lecture 15 - Proof of Convergence of Perceptron Learning Algorithm


Lecture 16 - Deep Learning (CS7015): Linearly Separable Boolean Functions


Lecture 17 - Deep Learning (CS7015): Representation Power of a Network of Perceptrons


Lecture 18 - Deep Learning (CS7015): Sigmoid Neuron


Lecture 19 - Deep Learning (CS7015): A typical Supervised Machine Learning Setup


Lecture 20 - Deep Learning (CS7015): Learning Parameters: (Infeasible) guess work


Lecture 21 - Deep Learning (CS7015): Learning Parameters: Gradient Descent


Lecture 22 - Deep Learning (CS7015): Representation Power of Multilayer Network of Sigmoid Neurons


Lecture 23 - Feedforward Neural Networks (a.k.a multilayered network of neurons)


Lecture 24 - Learning Paramters of Feedforward Neural Networks (Intuition)


Lecture 25 - Output functions and Loss functions


Lecture 26 - Backpropagation (Intuition)


Lecture 27 - Backpropagation: Computing Gradients w.r.t. the Output Units


Lecture 28 - Backpropagation: Computing Gradients w.r.t. Hidden Units


Lecture 29 - Backpropagation: Computing Gradients w.r.t. Parameters


Lecture 30 - Backpropagation: Pseudo code


Lecture 31 - Derivative of the activation function


Lecture 32 - Information content, Entropy and cross entropy


Lecture 33 - Recap: Learning Parameters: Guess Work, Gradient Descent


Lecture 34 - Contours Maps


Lecture 35 - Momentum based Gradient Descent


Lecture 36 - Nesterov Accelerated Gradient Descent


Lecture 37 - Stochastic And Mini-Batch Gradient Descent


Lecture 38 - Tips for Adjusting Learning Rate and Momentum


Lecture 39 - Line Search


Lecture 40 - Gradient Descent with Adaptive Learning Rate


Lecture 41 - Bias Correction in Adam


Lecture 42 - Eigenvalues and Eigenvectors


Lecture 43 - Linear Algebra : Basic Definitions


Lecture 44 - Eigenvalue Decompositon


Lecture 45 - Principal Component Analysis and its Interpretations


Lecture 46 - PCA: Interpretation 2


Lecture 47 - PCA: Interpretation 3


Lecture 48 - PCA: Interpretation 3 (Continued...)


Lecture 49 - PCA: Practical Example


Lecture 50 - Singular Value Decomposition


Lecture 51 - Introduction to Autoncoders


Lecture 52 - Link between PCA and Autoencoders


Lecture 53 - Regularization in autoencoders (Motivation)


Lecture 54 - Denoising Autoencoders


Lecture 55 - Sparse Autoencoders


Lecture 56 - Contractive Autoencoders


Lecture 57 - Bias and Variance


Lecture 58 - Train error vs Test error


Lecture 59 - Train error vs Test error (Recap)


Lecture 60 - True error and Model complexity


Lecture 61 - L2 regularization


Lecture 62 - Dataset augmentation


Lecture 63 - Parameter sharing and tying


Lecture 64 - Adding Noise to the inputs


Lecture 65 - Adding Noise to the outputs


Lecture 66 - Early stopping


Lecture 67 - Ensemble Methods


Lecture 68 - Dropout


Lecture 69 - A quick recap of training deep neural networks


Lecture 70 - Unsupervised pre-training


Lecture 71 - Better activation functions


Lecture 72 - Better initialization strategies


Lecture 73 - Batch Normalization


Lecture 74 - One-hot representations of words


Lecture 75 - Distributed Representations of words


Lecture 76 - SVD for learning word representations


Lecture 77 - SVD for learning word representations (Continued...)


Lecture 78 - Continuous bag of words model


Lecture 79 - Skip-gram model


Lecture 80 - Skip-gram model (Continued...)


Lecture 81 - Contrastive estimation


Lecture 82 - Hierarchical softmax


Lecture 83 - GloVe representations


Lecture 84 - Evaluating word representations


Lecture 85 - Relation between SVD and Word2Vec


Lecture 86 - The convolution operation


Lecture 87 - Relation between input size, output size and filter size


Lecture 88 - Convolutional Neural Networks


Lecture 89 - Convolutional Neural Networks (Continued...)


Lecture 90 - CNNs (success stories on ImageNet)


Lecture 91 - CNNs (success stories on ImageNet) (Continued...)


Lecture 92 - Image Classification continued (GoogLeNet and ResNet)


Lecture 93 - Visualizing patches which maximally activate a neuron


Lecture 94 - Visualizing filters of a CNN


Lecture 95 - Occlusion experiments


Lecture 96 - Finding influence of input pixels using backpropagation


Lecture 97 - Guided Backpropagation


Lecture 98 - Optimization over images


Lecture 99 - Create images from embeddings


Lecture 100 - Deep Dream


Lecture 101 - Deep Art


Lecture 102 - Fooling Deep Convolutional Neural Networks


Lecture 103 - Sequence Learning Problems


Lecture 104 - Recurrent Neural Networks


Lecture 105 - Backpropagation through time


Lecture 106 - The problem of Exploding and Vanishing Gradients


Lecture 107 - Some Gory Details


Lecture 108 - Selective Read, Selective Write, Selective Forget - The Whiteboard Analogy


Lecture 109 - Long Short Term Memory (LSTM) and Gated Recurrent Units (GRUs)


Lecture 110 - How LSTMs avoid the problem of vanishing gradients


Lecture 111 - How LSTMs avoid the problem of vanishing gradients (Continued...)


Lecture 112 - Introduction to Encoder Decoder Models


Lecture 113 - Applications of Encoder Decoder models


Lecture 114 - Attention Mechanism


Lecture 115 - Attention Mechanism (Continued...)


Lecture 116 - Attention over images


Lecture 117 - Hierarchical Attention