mxagar/deep_learning_udacity

These are my personal notes taken while following the Udacity Deep Learning Nanodegree.

Udacity Deep Learning Nanodegree: Personal Notes

These are my personal notes taken while following the Udacity Deep Learning Nanodegree.

The nanodegree is composed of six modules:

1. Introduction to Deep Learning
2. Neural Networks and Pytorch/Keras Guides
3. Convolutional Neural Networks (CNN)
4. Recurrent Neural Networks (RNN)
5. Generative Adversarial Networks (GAN)
6. Deploying a Model with AWS SageMaker

Additionally, I have added an extra module/subfolder which I will extend with new architectures, applications and tools that appeared post 2018: Extra.

Each module has a folder with its respective notes; you need to go to each module folder and follow the Markdown file in them.

Finally, note that:

• I have also notes on the Udacity Computer Vision Nanodegree in my repository computer_vision_udacity; that MOOC is strongly related and has complementary material.
• In addition to the Pytorch guide, I have a Keras guide; both condense the most important features of both frameworks. Currently, the Pytorch guide is more detailed.
• I have many hand-written notes you can check, too (see the PDFs).
• I have a specific repository for object detection and semantic segmentation, where additionally labeling tools are covered: detection_segmentation_pytorch.
• The exercises are commented in the Markdown files and linked to their location; most of the exercises are located in other repositories, originally forked from Udacity and extended/completed by me:
  • deep-learning-v2-pytorch
  • CVND_Exercises
  • DL_PyTorch
  • CVND_Localization_Exercises
  • sagemaker-deployment

Projects

Udacity requires the submission of a project for each module; these are the repositories of the projects I submitted:

1. Predicting Bike Sharing Patterns with Neural Networks Written from Scratch with Numpy: project-bikesharing.
2. Dog Breed Classification with Convolutional Neural Networks (CNNs) and Transfer Learning: project-dog-classification.
3. Text Generation: TV Script Creation with a Recurrent Neural Network (RNN): text_generator.
4. Face Generation with a Convolutional Generative Adversarial Network (GAN): face_generator_gan.
5. Sentiment Analysis RNN Deployed Using AWS SageMaker: sentiment_rnn_aws_deployment.

Practical Installation Notes

I basically followed the installation & setup guide from deep-learning-v2-pytorch, which can be summarized with the following commands:

# Create new conda environment to be used for the nanodegree
conda create -n dlnd python=3.6
conda activate dlnd
conda install pytorch torchvision -c pytorch
conda install pip

# Go to the folder where the Udacity DL exercises are cloned, after forking the original repo
cd ~/git_repositories/deep-learning-v2-pytorch
pip install -r requirements.txt

Mikel Sagardia, 2022.
No guarantees.

If you find this repository helpful and use it, please link to the original source.