For the Hour of Code / Hausner Code Fest, 6th to 8th graders.
Page easy url: http://bit.ly/code-fest
Perceptron model, Feed-forward Neural Networks
Handwritten digit recognition demo: https://transcranial.github.io/keras-js/#/mnist-cnn
See the learning and optimization in action: http://cs.stanford.edu/people/karpathy/convnetjs/
(Notebook) https://github.com/ea167/code-fest/blob/master/notebooks/1_Simple-DNN-for-MNIST.ipynb
2 basic layers
5 layers: 2 convolutional + 1 to group ("max pooling") + 2 layers
Demo Inception v3: https://transcranial.github.io/keras-js/#/inception-v3
If you feel doubtful, try these URLs as images :
- https://static.boredpanda.com/blog/wp-content/uploads/2017/09/funny-dog-bios-92-59ad3f9cd01a7__700.jpg
- https://static.boredpanda.com/blog/wp-content/uploads/2017/09/funny-dog-bios-70-59ad0f4a8264b__700.jpg
- https://static.boredpanda.com/blog/wp-content/uploads/2014/06/cute-bunnies-tongues-13.jpg
- https://images-eu.ssl-images-amazon.com/images/I/61qzr5q6l5L._SL1500_.jpg
Other pre-trained Networks: ResNet, SqueezeNet, VGG
Style Transfer: https://tenso.rs/demos/fast-neural-style/
Install Docker (PC or Mac): https://www.docker.com/community-edition
Launch Docker, and download my image for AI|DeepLearning:
docker pull ea167/jupyker-cpu
Then follow instructions here to launch: https://hub.docker.com/r/ea167/jupyker-cpu/~/dockerfile/
Or better if you have a PC with Nvidia GPU:
docker pull ea167/jupyker
and https://hub.docker.com/r/ea167/jupyker/~/dockerfile/
https://harishnarayanan.org/writing/artistic-style-transfer/ | Many images are borrowed from this great site.