Tile2Vec: Unsupervised representation learning for spatially distributed data

TL;DR

Tile2Vec is an unsupervised learning method that creates meaningful spatial data representations, improving classification and enabling analogies, inspired by word embedding techniques.

Contribution

It introduces Tile2Vec, adapting distributional hypothesis to spatial data for the first time, enhancing geospatial analysis with learned representations.

Findings

Improves downstream classification performance

Enables visual analogies in spatial data

Learns semantically meaningful representations

Abstract

Geospatial analysis lacks methods like the word vector representations and pre-trained networks that significantly boost performance across a wide range of natural language and computer vision tasks. To fill this gap, we introduce Tile2Vec, an unsupervised representation learning algorithm that extends the distributional hypothesis from natural language -- words appearing in similar contexts tend to have similar meanings -- to spatially distributed data. We demonstrate empirically that Tile2Vec learns semantically meaningful representations on three datasets. Our learned representations significantly improve performance in downstream classification tasks and, similar to word vectors, visual analogies can be obtained via simple arithmetic in the latent space.