Neural Network Classifiers

Building Neural Network Classifiers for Text and Images

This notebook shows how build neural networks using Scikit-learn and Keras to classify text and images.

The Data

• Part 1 of this project uses categorized-comments.jsonl:
  • Contains 450,000 rows with two columns- category (cat) and text (txt)
• Part 2 of this project uses the MNIST dataset included in Keras:
  • You can import the data directly into testing and training sets using this code:

    from keras.datasets import mnist
    (X_train, y_train), (X_test, y_test) = mnist.load_data()
    

    Tasks

Part 1: Text Classification

1. Build a neural network with Scikit-learn

2. Build a neural network with Keras

3. Use the neural networks to classify Reddit comments into one of four categories:

   a. News

   b. Science and Technology

   c. Sports

   d. Video Games

4. Report each model’s accuracy, precision, recall, F1, and confusion matrix

Part 2: Image Classification

1. Load the MNIST dataset from Keras into testing and training sets

2. Reshape and rescale the images to feed into the model

3. Build a neural network with Keras

4. Train and test the model

5. Report the model’s accuracy and loss

Results

Below, you can find the accuracy for each model, but you can find the remainder of the evaluation metrics in the Jupyter Notebook located in this project’s GitHub repository.

Text Classification

Scikit-learn Accuracy: 0.61725

Keras Accuracy: 0.64225

Image Classification

Keras Accuracy: 0.9855999946594238

Author

Xander Hieken