Course material
Dog vs cat dataset (used in most ipython notebooks)
| Background | 78 min | |||||
|---|---|---|---|---|---|---|
| Background | pdf key | 2 min | ||||
| Linear Algebra and Gradients | pdf key | 19 min | ||||
| Probability, Likelihood, Sampling, and Expectation | pdf key | 16 min | ||||
| Tensors | pdf key | 5 min | ||||
| Data as Tensors | ipynb html | 9 min | ||||
| Broadcasting | ipynb html | 11 min | ||||
| Setup and Tensorboard | ipynb html | 15 min | ||||
| Summary | pdf key | 2 min | ||||
| First Example | 107 min | |||||
| Linear Classifiers and Regression | pdf key | 7 min | ||||
| Linear Regression | pdf key | 6 min | ||||
| Linear Classification | pdf key | 8 min | ||||
| Linear Classification in Action | ipynb html | 13 min | ||||
| Linear Multi-class Classification | pdf key | 11 min | ||||
| Optimization and Gradient Descent | pdf key | 12 min | ||||
| Gradient Descent in Action | ipynb html | 10 min | ||||
| Computation Graphs | pdf key | 6 min | ||||
| Building a Computation Graph | ipynb html | 10 min | ||||
| Gradient Computation on Computation Graphs | pdf key | 10 min | ||||
| Back Propagation in Action | ipynb html | 12 min | ||||
| Summary | pdf key | 2 min | ||||
| Deep Networks | 135 min | |||||
| Beyond Linear Models | pdf key | 2 min | ||||
| Limitations of Linear Models | pdf key | 9 min | ||||
| Non-linearities (ReLU) | pdf key | 11 min | ||||
| Output Representations | pdf key | 9 min | ||||
| Loss Functions | pdf key | 8 min | ||||
| Building a Deep Network in PyTorch | ipynb html | 15 min | ||||
| Optimization of Deep Networks | pdf key | 6 min | ||||
| Stochastic Gradient Descent | pdf key | 11 min | ||||
| Mini-batches | pdf key | 7 min | ||||
| Momentum | pdf key | 7 min | ||||
| Optimization in PyTorch | ipynb html | 20 min | ||||
| What Is a Layer? | pdf key | 6 min | ||||
| Activation Functions | pdf key | 15 min | ||||
| Hyper-parameters | pdf key | 2 min | ||||
| Summary, a Practical Guide to Deep Network Design | pdf key | 7 min | ||||
| Convolutional Networks | 172 min | |||||
| Images and Structure | pdf key | 4 min | ||||
| High Dimensional Inputs | pdf key | 4 min | ||||
| Convolutions | pdf key | 22 min | ||||
| Convolutional Network in PyTorch | ipynb html | 8 min | ||||
| Convolutional Operators and Their Structure | pdf key | 21 min | ||||
| Average Pooling | pdf key | 6 min | ||||
| Max Pooling | pdf key | 6 min | ||||
| Convolutional Operations in PyTorch | ipynb html | 7 min | ||||
| Receptive Fields | pdf key | 12 min | ||||
| Design Principles of Convolutional Networks | pdf key | 16 min | ||||
| Building Efficient Convolutional Networks | ipynb html | 13 min | ||||
| Deep Representations and Exploiting the Structure of the Data | pdf key | 10 min | ||||
| Examining the Structure of Deep Networks | ipynb html | 15 min | ||||
| Dilation | pdf key | 9 min | ||||
| Up-Convolution | pdf key | 11 min | ||||
| Summary | pdf key | 5 min | ||||
| Making It Work | 307 min | |||||
| Practical Deep Learning | pdf key | 7 min | ||||
| Looking at Your Data | pdf key | 6 min | ||||
| Training, Validation, and Test Sets | pdf key | 14 min | ||||
| Distribution of Data | ipynb html | 12 min | ||||
| Network Initialization | pdf key | 12 min | ||||
| Random Initialization | pdf key | 9 min | ||||
| Xavier and Kaiming Initialization | pdf key | 20 min | ||||
| Initialization in PyTorch | ipynb html | 5 min | ||||
| Optimization | pdf key | 2 min | ||||
| Input Normalization | pdf key | 12 min | ||||
| Vanishing and Exploding Gradients | pdf key | 12 min | ||||
| Normalization | pdf key | 2 min | ||||
| Batch Normalization | pdf key | 9 min | ||||
| Layer Normalization | pdf key | 3 min | ||||
| Instance Normalization | pdf key | 3 min | ||||
| Group Normalization and Local Response Normalization | pdf key | 6 min | ||||
| Where to Add Normalizations? | pdf key | 8 min | ||||
| Normalizations in PyTorch | ipynb html | 5 min | ||||
| Residual Connections | pdf key | 13 min | ||||
| Residual Connections in Practice | ipynb html | 8 min | ||||
| Optimization Algorithms | pdf key | 10 min | ||||
| Learning Rate | pdf key | 9 min | ||||
| Learning Rate Schedules in PyTorch | ipynb html | 10 min | ||||
| Open Problem: Pruning and Compression | pdf key | 12 min | ||||
| Overfitting and How to Detect It | pdf key | 7 min | ||||
| Early Stopping | pdf key | 3 min | ||||
| Data Augmentation | pdf key | 12 min | ||||
| Dropout | pdf key | 11 min | ||||
| Weight Decay | pdf key | 6 min | ||||
| Ensembles | pdf key | 8 min | ||||
| Reducing Overfitting | repo | 20 min | ||||
| Transfer Learning | pdf key | 9 min | ||||
| Open Problem: Understanding Generalization | pdf key | 13 min | ||||
| Summary, a Practical Guide to Deep Network Optimization | pdf key | 11 min | ||||
| Computer Vision | 254 min | |||||
| Computer Vision Tasks | pdf key | 15 min | ||||
| Image Classification | pdf key | 13 min | ||||
| Case Study: AlexNet | pdf key | 19 min | ||||
| Case Study: VGG | pdf key | 11 min | ||||
| 1x1 Convolutions and Factorization | pdf key | 7 min | ||||
| Case Study: Network in Network | pdf key | 3 min | ||||
| Case Study: Inception Architecture | pdf key | 8 min | ||||
| Case Study: Residual Networks | pdf key | 11 min | ||||
| Factorization and Light-weight Networks | pdf key | 5 min | ||||
| Case Study: MobileNet | pdf key | 5 min | ||||
| Using Pre-trained Architectures | ipynb html | 11 min | ||||
| Object Detection | pdf key | 11 min | ||||
| Case Study: RCNN | pdf key | 13 min | ||||
| Case Study: Faster RCNN | pdf key | 14 min | ||||
| Case Study: RetinaNet | pdf key | 10 min | ||||
| Segmentation | pdf key | 10 min | ||||
| Case Study: FCN | pdf key | 11 min | ||||
| Case Study: Dilated Convolutional Networks | pdf key | 10 min | ||||
| Case Study: Mask RCNN | pdf key | 9 min | ||||
| Open Problem: Object Representations | pdf key | 12 min | ||||
| Temporal Models | pdf key | 6 min | ||||
| 3D Convolutions | pdf key | 5 min | ||||
| 2+1D Convolutions | pdf key | 3 min | ||||
| Case Study: I3D | pdf key | 6 min | ||||
| Open Problem: Effective Temporal Operations | pdf key | 10 min | ||||
| Open Problem: What Should We Infer or Label? | pdf key | 12 min | ||||
| Summary | pdf key | 4 min | ||||
| Reinforcement Learning | 210 min | |||||
| Acting in an Environment | pdf key | 23 min | ||||
| Acting in SuperTuxKart | ipynb html | 17 min | ||||
| Imitation Learning | pdf key | 11 min | ||||
| Dagger | pdf key | 8 min | ||||
| Dagger vs Imitation Learning | ipynb html | 30 min | ||||
| Non-differentiability | pdf key | 13 min | ||||
| REINFORCE | pdf key | 11 min | ||||
| Implementing REINFORCE | ipynb html | 30 min | ||||
| Policy Gradient | pdf key | 20 min | ||||
| Gradient Free Optimization | pdf key | 17 min | ||||
| Gradient Free Optimization in PyTorch | ipynb html | 13 min | ||||
| Open Problem: Structure Vs Data | pdf key | 13 min | ||||
| Summary | pdf key | 5 min | ||||
| Sequence Modeling | 131 min | |||||
| Sequence Models | pdf key | 6 min | ||||
| Recurrent Neural Networks | pdf key | 16 min | ||||
| Training Recurrent Networks | pdf key | 14 min | ||||
| LSTMs and GRUs | pdf key | 15 min | ||||
| Temporal Convolutions | pdf key | 16 min | ||||
| Sampling in Sequence Models | pdf key | 10 min | ||||
| Case Study: WaveNet | pdf key | 12 min | ||||
| Sequence Models in PyTorch | ipynb html | 20 min | ||||
| Attention and Transformers | pdf key | 16 min | ||||
| Summary | pdf key | 6 min | ||||
| Special Topics | 156 min | |||||
| Embedding Learning | 38 min | |||||
| Learning with an Expanding Set of Labels | pdf key | 4 min | ||||
| Embedding Learning | pdf key | 7 min | ||||
| Contrastive Losses | pdf key | 8 min | ||||
| Triplet Losses | pdf key | 5 min | ||||
| Selecting Training Examples | pdf key | 11 min | ||||
| Summary | pdf key | 4 min | ||||
| Generative Models | 80 min | |||||
| Image Generation | pdf key | 3 min | ||||
| Autoencoders | pdf key | 9 min | ||||
| Variational Autoencoders | pdf key | 12 min | ||||
| Transforming Noise | pdf key | 4 min | ||||
| Generative Adversarial Networks | pdf key | 11 min | ||||
| Pix2Pix | pdf key | 5 min | ||||
| CycleGan | pdf key | 10 min | ||||
| Image Editing | pdf key | 5 min | ||||
| Style Transfer | pdf key | 11 min | ||||
| Open Problem: Understanding Generative Models and Invariances | pdf key | 8 min | ||||
| Summary | pdf key | 2 min | ||||
| Adversarial Attacks | 38 min | |||||
| Fooling Deep Networks | pdf key | 5 min | ||||
| Finding Adversarial Examples | pdf key | 11 min | ||||
| Defense Through Data Augmentation | pdf key | 5 min | ||||
| White vs Black Box Attacks | pdf key | 7 min | ||||
| Open Problem: Realistic Attacks and Defenses | pdf key | 9 min | ||||
| Summary | pdf key | 2 min | ||||
| Final Words | 39 min | |||||
| Open Problem: Bias, Fairness, and Ethics in Deep Learning | pdf key | 29 min | ||||
| Course Summary and Further Topics | pdf key | 10 min | ||||