/advanced-tensorflow

Little More Advanced TensorFlow Implementations

Primary LanguageJupyter Notebook

Advanced TensorFlow

Collection of (Little More + Refactored) Advanced TensorFlow Implementations. Try my best to implement algorithms with a single Jupyter Notebook.

AutoEncoder

  • Denoising AutoEncoder
  • Convolutional AutoEncoder (using deconvolution)
  • Variational AutoEncoder

Adversarial Variational Bayes

  • AVB on 2-dimensional Toy Example

Basics

  • Basic Classification (MLP and CNN)
  • Custom Dataset Generation
  • Classification (MLP and CNN) using Custom Dataset
  • OOP Style Implementation of MLP and CNN

Class Activation Map

  • Pretrained Network Usage with TF-SLIM
  • Class Activation Map with Pretrained Network

Char-RNN

  • Preprocess Linux Kernel Sources
  • Train and Sample with Char-RNN

Domain Adaptation

  • Domain Adversarial Neural Network with Gradient Reversal Layer

Generative Adversarial Network

  • Deep Convolutional Generative Adversarial Network with MNIST

Mixture Density Network

  • Mixture Density Network
  • Heteroscedastic Mixture Density Network

Reinforcement Learning

  • Model Based RL (Value Iteration and Policy Iteration)

TF-SLIM

  • MNIST Classification with TF-SLIM

Super Resolution

  • Super-resolution with Generative Adversarial Network

Requirements

  • Python-2.7
  • TensorFlow-1.0.1
  • SciPy
  • MatplotLib
  • Jupyter Notebook