Assignments of the Udacity MOOC Deep Learning
This repository stores the assignments I have done during the Udacity MOOC on Deep Learning. The course is free but there is no support expect if you register to a nanodegree program. That's why I have I tried to comment as much as possible my results, you may find them useful.
Disclaimer: there is probably a code of conduct if you are enrolled in the nanodegree. Copying the code from the repository and pretending it to be sure would be bad. Even looking into it may be an infringement. Talk to your instructor.
Note that 4 GB of RAM are short to run the notebooks, 8 GB will be more comfortable.
I already had Jupyter Notebook running on my Mac. For reference, it has been shipped with Anaconda which is my Python bundle of choice. It works on Linux as well.
The only module to install from there is TensorFlow. The installation from pip is fine to me, the only trick is that the version 0.6.0 has been required (not the latest one). You can refer to the official documentation. So the installation is done with a single command line:
pip install --upgrade https://storage.googleapis.com/tensorflow/mac/tensorflow-0.5.0-py2-none-any.whl
You can now grab the assignments from here to run my code or the stubs from the TensorFlow repository.
From the repository root, start Notebook with jupyter notebook.
This repository is mainly for my own purpose but I would be happy to share some thoughts with you. There are probably many ways to improve and tune the various neural networks.
To do so, just fork this repository.
- all the assignments come from Udacity and the TensorFlow repository
- the weird fonts data set used for this training is the notMNIST dataset from Yaroslav Bulatov (there are also a download mirrors for the large and small dataset)
- the final neural network of the assignment #4 is loosely inspired by the notorious LeNet-5, as described in the research paper "Gradient-Based Learning Applied to Document Recognition" by Y. LeCun, L. Bottou, Y. Bengio and P. Haffner