Template Credit: Adapted from a template made available by Dr. Jason Brownlee of Machine Learning Mastery.

SUMMARY: The purpose of this project is to construct a predictive model using various machine learning algorithms and to document the end-to-end steps using a template. Jason Brownlee of Machine Learning Mastery posted a tutorial on deep learning using Kaggle’s Cats and Dogs dataset.

Jason’s script utilized the stochastic gradient descent and momentum (SGD) optimizer for his experiments. This script will replicate the same experiments with the Adam optimizer (with a learning rate of 0.001), so we can observe how different optimizers affect the model’s performance.

ANALYSIS: For the one block VGG model, the SGD optimizer achieved an accuracy of 72.331% after 20 epochs. The Adam optimizer achieved an accuracy of 69.253% using the same number of epochs.

For the two-block VGG model, the SGD optimizer achieved an accuracy of 76.646% after 20 epochs. The Adam optimizer achieved an accuracy of 71.759% using the same number of epochs.

For the three-block VGG model, the SGD optimizer achieved an accuracy of 80.184% after 20 epochs. The Adam optimizer achieved an accuracy of 73.870% using the same number of epochs.

For the VGG-3 with Dropout (0.2, 0.2, 0.2, 0.5) model, the SGD optimizer achieved an accuracy of 81.279% after 50 epochs. The Adam optimizer achieved an accuracy of 84.769% after the same number of epochs. Furthermore, another VGG-3 with Dropout (0.2, 0.3, 0.4, 0.5) model achieved an accuracy of 85.118% using Adam.

For the VGG-3 with image data augmentation model, the SGD optimizer achieved an accuracy of 85.816% after 50 epochs. The Adam optimizer achieved an accuracy of 91.449% after the same number of epochs.

For the VGG-3 with Dropout (0.2, 0.3, 0.4, 0.5) and image data augmentation model, the Adam optimizer achieved an accuracy of 90.227% after 50 epochs.

Dataset Used: Cats and Dogs Dataset

Dataset ML Model: Binary classification with numerical attributes

Dataset Reference: https://www.microsoft.com/en-us/download/details.aspx?id=54765

One potential source of performance benchmarks: https://www.kaggle.com/c/dogs-vs-cats/overview

The HTML formatted report can be found here on GitHub.