Autoencoders Flashcards
What are autoencoders?
Autoencoders are a type of supervised ANN with the goal of learning a representation of the input data in such a way that it captures the essential features.
What is the primary property of an autoencoder?
Identical input and output size. An autoencoder should take an input and reconstruct it fully.
What is the ‘encoder’ in an autoencoder?
The encoder compresses the input data into a lower-dimensional representation.
What is the ‘decoder’ in an autoencoder?
The decoder takes the compressed representation from the encoder and attempts to reconstruct the original input.
What are the primary uses of autoencoders?
Autoencoders can be used for many tasks, including dimensionality reduction, image denoising or simple compression and decompression.
What is latent space?
Latent space is the compressed representation learned by the encoder.
What is the objective function?
The objective function is a function used to minimize the difference between input data and its reconstruction, which could be achieved using a loss function.
What is the curse of dimensionality?
The curse of dimensionality is that as you add more features to the dataset, the data gets more sparse, and is more prone to overfitting.
How can we fix the curse of dimensionality using autoencoders?
Autoencoders can be used to compress the data down into smaller dimensions, finding representations that still retain the relationships in the data.
