Multiple image retrieval algorithms:

1. K nearest neighbour (KNN)
2. Locality-sensitive hashing (LSH)
3. Deep stacked convolutional autoencoder (DCAE)
4. Deep siamese convolutional autoencoder with adaptive euclidean loss (SDCAE)
5. Winner takes all sparse convolutional autoencoder (WTA_CONV)

Data class - Allows for the data to be passed like the tensorflow MNIST data.

Metric fusion - Fuse multiple features together with minimal information loss