Autoencoding with K-Means
The MNIST dataset contains images of handwritten digits from 0 to 9. Using the machine learning K-Means algorithm to cluster the MNIST dataset achieves poor results. This experiment seeks to improve and measure the accuracy by using an autoencoder to massage the input data prior to K-Means. To measure the results a purity score will be generated and a confusion matrix will display the distribution of the clusters. The notebook is available on GitHub https://github.com/MichaelSchlachter/clustering/blob/master/mnist_encoder_kmeans.ipynb Generate…
0 Comments
July 12, 2020