MEMORY FACILITATION USING DIRECTED ACYCLIC GRAPHS

摘要

Memory facilitation using directed acyclic graphs is described, for example, where a plurality of directed acyclic graphs are trained for gesture recognition from human skeletal data, or to estimate human body joint positions from depth images for gesture detection. In various examples directed acyclic graphs are grown during training using a training objective which takes into account both connection patterns between nodes and split function parameter values. For example, a layer of child nodes is grown and connected to a parent layer of nodes using an initialization strategy. In examples, various local search processes are used to find good combinations of connection patterns and split function parameters.

主权项

1. A computer implemented method at a machine learning system comprising:
accessing a plurality of training examples; storing a graph of connected nodes at a memory of the machine learning system the nodes having split function parameters specifying tests for determining flow of elements through the graph; and adding, using a processor, new nodes to the graph in a manner which selects between possible connection patterns between nodes of the graph, and possible split function parameter values, by applying a training objective to elements of the training examples reaching nodes of the graph according to the split function parameters.