Deep Learning for Data Science

Tutorials on Deep Learning for Data Science with

⚠️ This repository is still WIP so content is still in draft mode - and references to external resources might be missing ⚠️

Content at a glance

I: ANN and Automatic Differentiation

1. Intro to Artificial Neural Networks

   • Short intro: Supervised vs Unsupervied Learning

   • Perceptron: the linear Neuron model

     • Short on Vectorisation
     • ADAptive LInear NEuron (ADALINE)

   • Multi-Layer Perceptron

     • numpy-based implementation
     • torch.Tensor-based implementation

   • From ANN to DNN

     • Introduction to torch.nn
     • PyTorch Model Persistence
     • Classification and Regression Revisited
     • Short on Universal Approximation Theorem
     • from Logistic to Softmax
     • Multi-class Classification and CrossEntropyLoss

2. Automatic Differentiation and autograd:

• Intro to Automatic Differentiation

  • forwad mode AD
  • backward mode AD
  • tangent and autograd

• Towards torch.nn: micrograd

  • torch.Tensor and autograd

II: Data and Dataset

3. Data for Machine and Deep Learning

• Data for Machine (Deep) Learning

  • torchvision
  • torchtext
  • torchaudio

• Deep learning for Data

  • Choose your Estimator
  • Choose your DL model

• Data the torch way - Introducing torch.utils.data, DataSet, and DataLoader

  • Preparing Data for Experiments - Training, Test and Cross Validation

Requirements

This tutorial runs on Python 3 (Py3.7+ should be fine), and requires the following main packages:

• numpy
• scipy
• matplotlib
• scikit-learn
• pandas
• torch (of course 😄)
• torchvision

The complete list of requirements is available in requirements.txt

Detailed (step-by-step) instructions to setup the Python virtual environment on your local machine are also available here.

License Summary

The material provided in this repository adopts two different licence files, for Lecture notes and Source Code, respectively.

The Lecture notes (and corresponding source notebooks) are available under the Creative Commons Attribution-ShareAlike 4.0 International License.

The samples and reference code within this repository is made available under the Apache License 2.0. See the LICENSE file.

References

Author: Valerio Maggio, Senior Research Associate @ Dynamic Genetics Lab

Contact
@leriomaggio
ValerioMaggio
valerio.maggio@bristol.ac.uk