Deep Learning Examples
This is a repository of some examples in Deep Learning implemented in PyTorch/TensorFlow.
Contact info: yuxuan.chen@fu-berlin.de
Table of Content
1. Basic
1.1 Image Classification with Logistic Regression
Logistic regression on a subset of MNIST dataset containing 20000 samples. Implemented in:
- PyTorch: 91.1% validation accuracy
- TensorFlow: 90.8% validation accuracy
1.2 Image Classification with Neural Network
A simple neural network of one hidden layer on the same dataset. Implemented in:
- PyTorch: 96.9% validation accuracy
- TensorFlow 96.4% validation accuracy
1.3 Image Classification with CNN
CNN of 2 convolutional layers on the same dataset. Implemented in:
- PyTorch: 98.9% validation accuracy
- TensorFlow: 97.6% validation accuracy
2. Intermediate
2.1 Sequence Classification with RNN
Use RNN and its variations to classify DNA sequences. Implemented in:
- PyTorch: 100% validation accuracy
- TensorFlow: 100% validation accuracy
2.2 VAE
We use VAE and CVAE of 2-dim latent space to generate pictures learnt from 20000-sized MNIST dataset. Implemented in:
2.3 GAN
We use vanilla GAN of 32-dim latent space to generate from 60000-sized MNIST dataset. Implemented in:
3. Advanced
Dependencies
- Python 3.5+
- PyTorch 1.5.0
- TensorFlow 2.2.0