Skip to content

Latest commit

 

History

History
27 lines (23 loc) · 929 Bytes

README.md

File metadata and controls

27 lines (23 loc) · 929 Bytes

autoencoder-showcase

Welcome to the supplementary experiments for the paper 'A Showcase of the Use of Autoencoders in Feature Learning Applications'

Install this package using:

devtools::install_github("ari-dasci/autoencoder-showcase")

Inside this package, you will find 4 main functions:

  • anomaly_detection()
    Creates a synthetic multi-valued time series with an anomalous region and performs anomaly detection.
  • visualization()
    Downloads the Statlog dataset and compacts it to 2 and 3 dimensions for visualization.
  • hashing()
    Loads IMDB dataset from Keras, trains an autoencoder and hashes the test subset, measuring the correspondance between distance among instances and Hamming distance among their hashes.
  • denoising()
    Loads CIFAR10 dataset and trains a denoising autoencoder, performs denoising over the test subset.