Confidence estimation for t-SNE embeddings using random forest

Abstract Dimensionality reduction algorithms are commonly used for reducing the dimension of multi-dimensional data to visualize them on a standard display. Although many dimensionality such as t-distributed Stochastic Neighborhood Embedding aim preserve close neighborhoods in low-dimensional space, they might not accomplish that every sample and eventually produce erroneous representations. In this study, we developed supervised confidence estimation algorithm detecting samples embeddings. Our generates score each an embedding based distance-oriented random forest regressor. We evaluate its performance both intra- inter-domain compare it with neighborhood preservation ratio our baseline. results showed resulting provides distinctive information about correctness any compared The source code is available at https://github.com/gsaygili/dimred .