paraDime: A Framework for Parametric Dimensionality Reduction

paraDime is a modular framework for specifying and training parametric dimensionality reduction (DR) models. These models allow you to add new data points to existing low-dimensional representations of high-dimensional data. ParaDime DR models are constructed from simple building blocks (such as Relations and Relation Transforms), so that experimentation with novel DR techniques becomes easy.

Here you can see a parametric version of t-SNE 1 trained on a subset of 5000 images of handwritten digits from the MNIST dataset 2:

Parametric t-SNE of a subset of the MNIST image dataset

The rest of the 60,000 images can then be easily embedded into the same space without retraining the t-SNE:

Remaining MNIST data embedded into the existing low-dimensional space

References

1

Van Der Maaten, L., Hinton, G. “Visualizing data using t-SNE”, Journal of Machine Learning Research (2008).

2

LeCun, Y., Cortes, C., Burges, C.J.C. “The MNIST database of handwritten digits” (1998).