FINAL PROJECT:Structures and Arguments > Deep Learning in PyTorch
R, this looks great. May I make a suggestion as demonstration projects? Look at Janelle Shane's work here:
http://aiweirdness.com/aboutme
Here is her archive of trainings:
http://aiweirdness.com/archive
My little pony names
craft bear names
knitting directions
Enjoy! Also, you will have used a "woman in comp sci/data sci" source, which is one way to build justice in the world.
http://aiweirdness.com/aboutme
Here is her archive of trainings:
http://aiweirdness.com/archive
My little pony names
craft bear names
knitting directions
Enjoy! Also, you will have used a "woman in comp sci/data sci" source, which is one way to build justice in the world.
May 7, 2018 |
MbS
Middle: We’ll cover a neural network used for handwritten digit classification (LeNet), and elaborate on the basic building blocks for neural networks (convolutional layers, fully connected layers, activation functions, backpropogation). Some advanced topics will also be covered, (Dropout/intuition behind ReLu activation function), to get our hands dirty with some of the state of the art additions to the field.
End: With the information presented in the middle, we’ll build our very own neural network and test it out! We’ll use the concepts covered in the body of the project when implementing our network to classify handwritten digits.
LOGOS:
1) Ideas are referenced from highly cited papers in the deep learning community
2) To implement our neural network, we’ll be using PyTorch, using the classic example of handwritten digits
PATHOS: Go into benefits of neural networks, the innovations it has generated in the world
ETHOS: We’ll be using some papers from the best researchers in the field (LeCun, Hinton, Bengio), Pytorch has 15,000+ stars on github
STRUCTURE: I will use two structures: 1) Informational Guide and 2) Illustrative Example (Neural Network in Pytorch!)
GOAL: Guide a newcomer in the field of deep learning with some of the tools and concepts that are used in the field