kevinlemon / finch Goto Github PK

• These are the models studied, implemented and tested by me in the past;
• I used lightweight datasets, so even if you don't have GPU, you can still run these scripts;
• If you would like to support, please Star the project, many thanks!

Guide

1. The following command clones all the files (>300MB);

git clone https://github.com/zhedongzheng/finch.git

2. Use contents to find the model and test that may interest you, click on that test

3. Find the test file path

4. run on command line

cd finch/nlp-models/tensorflow
python rnn_attn_estimator_imdb_test.py

Library

Py3 is perferred, but Py2 should also work in theory (if it doesn't please raise an issue)

• tensorflow >= 1.4
• scikit-learn
• openai-gym
• nltk
• opencv 3

Style

• Model is implemented very early, using feed_dict (most common but slowest data pipeline);
• I am moving towards Estimator, which is based on tf.estimator.Estimator, more efficient;

Contents

• NLP
  • Text Representation
  • Text Classification
  • Text Generation
  • Text Labelling
  • Text to Text
  • Image to Text
• Computer Vision
• Information Retrieval
• Reinforcement Learning
• Appendix
  • Shallow Structure Models
  • Ensemble
  • Time Series

NLP

Text Representation

• Python   |   LSA   |   Model for Visualization     Test     Result   |   Model for Concepts     Test     Result   |  

• Python   |   TF-IDF     Model     Test     Result   |  

• TensorFlow   |   Word Embedding - Skip-Gram     Model     Test   |  

• TensorFlow   |   Word Embedding - CBOW     Model     Test   |  

Text Classification

• Python   |   TF-IDF + LR     Model     IMDB Test   |  

• TensorFlow   |   CNN     Model     IMDB Test   |   Model (Multi-kernel)     IMDB Test     Result   |  

• TensorFlow   |   LSTM     Model     IMDB Test   |  

• TensorFlow   |   CNN-LSTM     Model     IMDB Test   |  

• TensorFlow   |   LSTM + Attention     Model     IMDB Test   |   Estimator     IMDB Test     IMDB Config   |  

Text Generation

• Python   |   2nd order Markov     Model     Robert Frost Test   |  

• TensorFlow   |   CNN-Highway-RNN     Paper     Model     PTB Test   |  

• TensorFlow   |   Char-RNN     Model   |   English Test     Chinese Test   |   Model (Beam-Search)     English Test   |  

• TensorFlow   |   Varational Recurrent Autoencoder   |  

Text Labelling

• TensorFlow   |   LSTM     Model   |   POS Tagging Test   |   Chinese Segmentation Test   |  

• TensorFlow   |   BiLSTM     Model   |   POS Tagging Test   |   Chinese Segmentation Test   |  

• TensorFlow   |   BiLSTM + CRF     Model   |   POS Tagging Test   |   Chinese Segmentation Test   |  

Text to Text

• TensorFlow   |   Seq2Seq     Model     Sorting Test   |   Estimator     Sorting Test   |  

  • TensorFlow   |   Seq2Seq + Attention     Model     Sorting Test   |  

  • TensorFlow   |   Seq2Seq + BiLSTM Encoder     Model     Sorting Test   |  

  • TensorFlow   |   Seq2Seq + Beam-Search     Model     Sorting Test   |  

  • TensorFlow   |   Seq2Seq + BiLSTM Encoder + Attention + Beam-Search     Model     Sorting Test   |  

• TensorFlow   |   Attention Is All You Need - Transformer   |  

Image To Text

(To run this section, you need to download COCO dataset first)

• TensorFlow   |   CNN + RNN     Model     COCO Test   |  

  • TensorFlow   |   CNN + RNN + Attention + Beam-Search     Model     COCO Test   |  

  • TensorFlow   |   Fine-tuning CNN + RNN + Attention + Beam-Search     Model     COCO Test   |  

Computer Vision

Image Classification

• Python   |   Bayesian Inference     Pixel Classification   |  

• TensorFlow   |   MLP     Model     MNIST Test     CIFAR10 Test   |  

• TensorFlow   |   CNN     Model     MNIST Test     CIFAR10 Test   |   Estimator     MNIST Test   |  

• TensorFlow   |   RNN     Model     MNIST Test     CIFAR10 Test   |   Estimator     MNIST Test   |  

Image Generation

• Autoencoder

  • TensorFlow   |   Stacked Autoencoder (weights-tied)     Model     MNIST Test   |  

  • TensorFlow   |   Denoising Autoencoder     Model     MNIST Test   |  

  • TensorFlow   |   Sparse Autoencoder     Model     MNIST Test   |  

  • TensorFlow   |   Variational Autoencoder     Model     MNIST Test   |  

  • TensorFlow   |   Conv2D Autoencoder (weights-tied)     Model     MNIST Test     CIFAR10 Test   |  

• Generative Adversarial Network

  • TensorFlow   |   DCGAN     Model     MNIST Test     Result   |   Conditional Model     MNIST Test   |  

OpenCV

• OP   |   Resize

• OP   |   Rotations

• Segmentation   |   Contours

• Segmentation   |   Sorting Contours

• Segmentation   |   Line detection

• Segmentation   |   Circle detection

• Segmentation   |   Blob detection

• Detection   |   Face & Eye Detection Using Cascade Classifier

• Detection   |   Walker & Car Detection Using Cascade Classifier

Information Retrieval

• Python   |   Apriori     Model     MovieLens Test   |  

• Python   |   Collborative Filtering   |   MovieLens     User-based Model     Test   |  

• TensorFlow   |   Matrix Factorization     Model     MovieLens Test   |  

Reinforcement Learning

• Python   |   Q-Learning     Model     CartPole Test   |  

• Python   |   Sarsa     Model     CartPole Test   |  

• TensorFlow   |   Policy Gradient     Model     CartPole Test   |  

Appendix

Shallow Structure Models

• Java   |   Logistic Regression     Model     Test   |  

• Java   |   Support Vector Machine     Model     Test   |  

• TensorFlow   |   Linear Regression     Model     Test   |  

• TensorFlow   |   Logistic Regression     Model     Test   |  

• TensorFlow   |   Support Vector Machine     Model     Test   |  

• TensorFlow   |   K Nearest Neighbors     Model     MNIST Test   |  

• TensorFlow   |   K-Means     Model     MNIST Test   |  

Ensemble

• Python   |   Bagging     Model     Test   |  

• Python   |   Adaboost     Pseudocode     Model     Test   |  

• Python   |   Random Forest     Model     Test   |  

• TensorFlow   |   Random Forest     Estimator & MNIST Test   |  

• TensorFlow   |   Gradient Boosting Trees     Estimator & MNIST Test   |  

Time Series

• TensorFlow   |   RNN     Model     Test     Preview   |