Generating a hierarchy of visual pattern classes

摘要

A hierarchy machine may be configured as a clustering machine that utilizes local feature embedding to organize visual patterns into nodes that each represent one or more visual patterns. These nodes may be arranged as a hierarchy in which a node may have a parent-child relationship with one or more other nodes. The hierarchy machine may implement a node splitting and tree-learning algorithm that includes hard-splitting of nodes and soft-assignment of nodes to perform error-bounded splitting of nodes into clusters. This may enable the hierarchy machine, which may form all or part of a visual pattern recognition system, to perform large-scale visual pattern recognition, such as font recognition or facial recognition, based on a learned error-bounded tree of visual patterns.

主权项

1. A method comprising:
classifying a reference set of visual patterns that belong to a parent class into mutually exclusive child classes that include first and second child classes, a visual pattern from the reference set being classified into the first child class instead of the second child class; modifying a weight vector that corresponds to the parent class, the modified weight vector altering a first probability that the visual pattern belongs to the first child class and a second probability that the visual pattern belongs to the second child class; based on the altered first and second probabilities, removing mutual exclusivity from the first and second child classes by adding the visual pattern to the second child class; and using a processor, generating a hierarchy of classes of visual patterns, the hierarchy including the parent class and the mutually nonexclusive first and second child classes that each include the visual pattern.