Back to Course
Deep Learning with PyTorch
Module 9 of 12
9. Autoencoders
1. Compression
Input -> [Encoder] -> Bottleneck -> [Decoder] -> Output. The goal: Output should equal Input. The Bottleneck forces the model to learn the "Latent Representation" (Essence) of the data.
2. Denoising
Input: Corrupted Image. Target: Clean Image. The model learns to remove noise (photoshop).