Objective: distinguish reconstruction, probabilistic generation and families of models.
An encoder turns x into a latent representation z; a decoder reconstructs
x̂. The network minimizes a reconstruction loss. If it has too much
capacity without constraint, it may learn a plain identity and produce a
useless representation.
x → encoder → z → decoder → x̂Uses: compression, denoising, pretraining, anomaly detection. A large reconstruction error is only a good anomaly score if the protocol confirms it.
The Variational Autoencoder learns a latent distribution and optimizes reconstruction plus KL regularization. Samples can be drawn in the latent space and then decoded. The trade-off between fidelity and latent structure is central.
Preview — the rest of the lesson is for enrolled readers.
Your access is tied to your account, not to this link. Sign in with the same email you used in class: your course is waiting, no need to enter the code again.
Sign inNo account yet? Create oneThe first modules of the course are open to everyone. For the rest you have three options: buy this course once and for all, subscribe, or enter the code handed out in class.