Tensorflow Applications

These applications are based on the information learnt in the TensorFlow in Practice Specialization.

Table of contents

• Tensorflow Applications
  • Table of contents
  • Prerequisites
  • Possible Issues
  • Convolutional Neural Networks
    • Basic Applications
    • Advanced Applications
  • Natural Language Processing
    • Basic Applications
    • Advanced Applications
      • IMDB Reviews Classification Comparison (Subwords)
      • IMDB Reviews Classification Comparison
      • News Headline Sarcasm Classification Comparison
  • Sequences, Time Series and Prediction
    • Basic Applications
    • Advanced Aplications

Prerequisites

• Some of the applications are using Tensorflow 2.0.0-rc1 and mostly others are using Tensorflow 1.14.0. You can create seperate conda environment for both version. I recommend gpu versions for fast training.

pip install tensorflow-gpu==1.14.0

pip install tensorflow-gpu==2.0.0-rc1

• Before you train in gpu you have to install and configure CuDNN, Cuda and Nvidia driver. For more information Tensorflow2.0.0 GPU Support

• Most of the application using specific dataset. Thanks to wget command, you don't need to effor for downloading them, just run the predetermined cell.

Possible Issues

• If you can't plotting the result just re-run the cell.

  <Figure size 640x480 with 1 Axes>
  <Figure size 640x480 with 1 Axes>

• If you install all of the gpu prerequisites, but still getting error:

    nvidia-smi

  Look-up the nvidia gpu memory usage, kill all of the process with kill [process-pid].

• If you worrying about some of the applications training time, don't worry if one epoch took less than 15 min. (Especcially for NLP applications)

Convolutional Neural Networks

Basic Applications

• Housing Price Prediction with Single Perceptron
• Fashion Recognition Using Deep Neural Network
• Handwritten Digit Recognition
• Horse-Human Classifier Using CNN
• Cat-Dog Classifier Using CNN (3k Images)
• Cat-Dog Classifier Using CNN (25k Images)
• GPU Training Test
• Multi Categorical Classification

Advanced Applications

• Image Augmentation
• Transfer Learning and Drop-out
• Hand Gesture Recognition

Natural Language Processing

Basic Applications

• Sentiment in Text
• Sentiment Preprocessing in BBC Text Archive
• Sentiment Preprocessing in Sarcasm Detection Dataset
• Text Classification Using Word Embeddings In IMDB Reviews Dataset
• Text Classification Using Word Embeddings In BBC News Archive
• Text Classification Using Word Embeddings In IMDB Reviews Dataset (Subwords8k)

Advanced Applications

• Exploring Overfitting in NLP
• Creating Shakespeare Sonnets Using LSTM

IMDB Reviews Classification Comparison (Subwords)

• IMDB Subwords 8K with 1D Convolutional Layer
• IMDB Subwords 8K with Single Layer LSTM
• IMDB Subwords 8K with Multi Layer LSTM

IMDB Reviews Classification Comparison

• IMDB Reviews Classification with 1D Convolutional Layer
• IMDB Reviews Classification with Bidirectional GRU
• IMDB Reviews Classification with Bidirectional LSTM

News Headline Sarcasm Classification Comparison

• Sarcasm Classification with 1D Convolutional Layer
• Sarcasm Classification with Bidirectional LSTM

Sequences, Time Series and Prediction

Basic Applications

• Creating Random Time Series
• Creating Dataset
• Naive Forecasting in Time Series
• Time Series Prediction with Single Layer Neural Network
• Time Series Prediction with DNN

Advanced Aplications

• Time Series Prediction with RNN
• Time Series Prediction with LSTM
• Time Series Prediction with CNN and LSTM
• Sunspot Prediction with CNN and LSTM