Catalogue of Artificial Intelligence Techniques


Jump to: Top | Entry | References | Comments

View Maths as: Images | MathML

Euclidean Embeddings

Keywords: euclidean, neighbour, vector

Categories: Speech

Author(s): Michael Clarke

Euclidean embeddings are a method of measuring ‘distances’ between data to group similar results and then run nearest neighbour type searches. Euclidean Embeddings are created by first mapping all the data in a 3D space according to properties of the results. Results with similar properties will be grouped together in certain dimensions of the map using Euclidean distances (distances along a straight line according to the displacement of each point). Result in similar locations can be mapped into groups (either multiple points in a single group or multiple points shared between multiple groups). If a point is then selected in one of these groups, other points in the group can also be used as alternatives. This is often used as a technique for bandwidth optimization by pre-grouping points to send and then sending an ID for the centre of the points. The Euclidean distances for all the other points within that group are then sent



Add Comment

No comments.