An introduction to PyTorch

This repository contains Jupyter notebooks that serve as an introduction to the basics of PyTorch. Each notebook will introduce different concepts as listed in the corresponding section.

Introduction to Deep Learning

Notebook dl-intro introduces basic concepts of DL using logistical regression.

What you will learn

1. Introduction to supervised learning and classification. Logistic regression using CIFAR10.
2. Using matplotlib.
3. Loss functions, activation functions and gradient descent.
4. PyTorch packages, tensors, and gradients.
5. Building neural network models using PyTorch

Fully connected networks

Notebook feedforward introduces basic layering architecture of neural networks using fully connected layers.

What you will learn

1. PyTorch datasets, dataloaders and data transforms
2. Data batches and stochastic gradient descent
3. Different activation functions, softmax and cross entropy loss
4. Using PyTorch on GPUs

Convolutional networks

Notebook convolution introduces convolutional layers

What you will learn

1. Convolution operation, kernels, strides.
2. Convolutional layers in PyTorch
3. Pooling and maxpooling in PyTorch
4. Using the confusion matrix to analyze results

Using Kaggle

Notebook detecting-emotions introduces the use of Kaggle and Kaggle datasets

What you will learn

1. How to use Kaggle and import Kaggle datasets from Colab or Kaggle itself
2. Reduce overfitting using dropout
3. Using PyTorch to load image datasets for classification

Custom Datasets

Notebook custom_dataset introduces building custom datasets in PyTorch

What you will learn

1. Using the pandas package
2. Building a custom dataset using PyTorch
3. Handling categorical data as input
4. Building multimodal models

Autoencoders

Notebook fraud-detection-pytorch introduces the autoencoders architectures in PyTorch and use it to detect credit card fraud

What you will learn

1. Autoencoders
2. Some functionality of sklearn
3. Some statistics