14.9 Applications of Autoencoders

Autoencoders have been successfully applied to dimensionality reduction and information retrieval.

Lower dimensional representations can improve performance on many tasks, such as classification.

Information retrieval: the task of finding entries in a database that resemble a query entry. Benefit:

• Usual benefits from dimensionality reduction that other tasks do
• search can be extremely efficient in certain kind of low dimensional spaces. Semantic hashing: train the D reduction algorithm to produce a code that is low-dimensional and binary, then we can store all database entries in a hash table that maps binary code vector to entries. We cam also search over slightly less similar entries efficiently, just by flipping individual bits from the encoding of the query.

Semantic hashing is applied to both textual input and image.