Deep Learning Tutorial - Lindsey Kitchell

This is a tutorial for combining Deep Learning and the Laplace Beltrami spectrum. This tutorial assumes some experience with python and machine learning. Experience using the machine learning modules from the python library sci-kit-learn will also be very helpful. The tutorial will be applicable beyond the Laplace Beltrami spectrum, but the data used will focus on it.

Start Here

Background information

I recommend to read the included documents in the following order:

1. Introduction to Deep Learning
2. Introduction to Keras
3. Multi-layer Perceptrons in Keras
4. Convolutional Neural Networks in Keras
5. Activation Functions
6. Other Useful Keras Functions
7. Loss Functions and Optimizers
8. Evaluating the Neural Networks (cross validation)
9. Data Preprocessing
10. Regularization
11. Hyperparameter Tuning

Working with data

If the links below will not load, please copy the URL and paste it into this site - https://nbviewer.jupyter.org/

12. MLP - Gender Classification
13. 1D CNN - Gender Classification
14. 2D CNN - Gender Classification
15. MLP - Multi-Class Classification
16. 1D CNN - Multi-Class Classification

Sources and Useful Resources

For learning about machine learning in general, I recommend this free introductory course from udacity. It will get you up and running with machine learning in python very quickly.

I also recommend this book by the creator of Keras (the python library used in this tutorial): Deep Learning with Python. It has a lot of detailed information about deep learning in general and how to implement it with Keras.

The following links were very helpful in the creation of this tutorial:

• https://www.youtube.com/watch?v=FrkYu2zVUyM
• https://github.com/leriomaggio/deep-learning-keras-tensorflow/tree/pydata-london2017
• https://machinelearningmastery.com/tutorial-first-neural-network-python-keras/
• https://elitedatascience.com/keras-tutorial-deep-learning-in-python#step-7
• https://github.com/keras-team/keras/tree/master/examples
• https://github.com/leriomaggio/deep-learning-keras-tensorflow/blob/euroscipy2017/1%20Multi-layer%20Fully%20Connected%20Networks.ipynb
• http://neuralnetworksanddeeplearning.com/chap1.html
• http://www.zhanjunlang.com/resources/tutorial/Deep%20Learning%20with%20Keras.pdf
• https://www.youtube.com/channel/UC9OeZkIwhzfv-_Cb7fCikLQ
• https://machinelearningmastery.com/applied-deep-learning-in-python-mini-course/
• https://github.com/fchollet/keras-resources