A project for developing numeric representations
of forest ecosystems using AI
Embeddings are numeric representations of abstract objects. Drawing from the discipline of Representation Learning, most of applications focus on the embedding of language or images. Latent Forest applies represenation approaches to forest ecosystems. This involves both trees and species embeddings, which capture characteristics such as habitat preference, tree size or phytosociology, and forest type embeddings, which represent ecosystem-level characteristics such as species composition, structure and productivity.