Loading your learn pack…
Read the summary, then start training!
The encoder maps each input x to parameters of a latent distribution qϕ(z|x) (often Gaussian), not to a single latent point.
In a variational autoencoder (VAE), the architecture is typically described as three connected parts: an encoder, a latent space, and a decoder. The encoder takes an input data point x (e.g., an image) and maps it into a probabilistic representation in the latent space, producing parameters of a variational distribution qϕ(z|x) rather than a single latent vector. This latent distribution is usually modeled as a multivariate Gaussian, so the encoder outputs values such as the mean Eϕ(x) and (optionally) a variance σϕ(x), defining how z is distributed for each input. The latent space is the probabilistic bottleneck where sampling occurs. During training, a latent variable z is sampled from qϕ(z|x) and then passed to the decoder. The decoder Dθ maps from the latent space back to the input space, producing a reconstruction x′ (or parameters of a likelihood distribution for x given z). In practice, the decoder is often implemented so that x|z is Gaussian with mean Dθ(z), and the reconstruction quality is measured with a reconstruction error (e.g., mean squared error or cross-entropy). Training the encoder and decoder jointly is done by maximizing a lower bound on the data likelihood, the ELBO. The ELBO combines (1) a reconstruction term that encourages x′ to match x and (2) a regularization term that makes the approximate posterior qϕ(z|x) align with the prior pθ(z), commonly via the Kullback–Leibler divergence. To enable gradient-based learning through the sampling step, VAEs use the reparameterization trick, rewriting the random sampling as a deterministic function of encoder outputs plus external noise.
The encoder maps each input x to parameters of a latent distribution qϕ(z|x) (often Gaussian), not to a single latent point.
The latent space acts as a probabilistic bottleneck: z is sampled from qϕ(z|x) and fed to the decoder.
The decoder maps z back to the input space to produce a reconstruction x′ (or likelihood parameters), and training uses the ELBO combining reconstruction error and a KL (or related) regularization term.
A neural network that maps input x to parameters of a variational posterior distribution qϕ(z|x) over latent variables z.
A low-dimensional probabilistic representation where each input corresponds to a distribution over latent variables z rather than a single point.
A neural network that maps sampled latent variables z to the input space, producing a reconstruction x′ or parameters of pθ(x|z).
The approximate distribution over latents produced by the encoder for a given input x.
A chosen distribution over latent variables (commonly N(0,I)) that regularizes the latent space.
The evidence lower bound objective that trades off reconstruction accuracy with regularization of qϕ(z|x) toward the prior pθ(z).
A method that rewrites sampling z~qϕ(z|x) as a differentiable transformation of encoder outputs plus external noise, enabling backpropagation.
“Can you explain what "The encoder maps each input x to parameters of a latent distribution qϕ(z|x) (often Gaussian), not to a single latent point." means in simple terms?”