Machine Learning method maps from $\mathbb{R}$ to $\mathbb{R}^2$

Question

I have worked with image classification and image segmentation etc.

While working with images we are either mapping from $\mathbb{R}^2\to\mathbb{R}$ or $\mathbb{R}^2\to\mathbb{R}^2$. Are there any machine leaning methods that can map from $\mathbb{R}\to\mathbb{R}^2$. For example, I have a feature vector of $3$-$5$ elements and I want to map it to $32\times32$ (approximately).

Is is possible? If it is, by which methods?

Answer 1

Sure. You can use a neural network with fully connected layers; or a neural network with deconvolutional layers. The latter is common in image segmentation tasks, so you may already be familiar with it.