Decision tree fields to map dataset content to a set of parameters

A labeling system comprising:

• one or more processors;
• a graph definer executed by the one or more processors and configured to define a factor graph in processor-readable memory, the factor graph including a factor graph model replicated for each variable node of a set of variable nodes, each variable node being informed by one or more factor types in the factor graph model, each factor type being implemented as a single decision tree, at least one factor type defining the relationship among more than two variable nodes;
• and
• a graph trainer executed by the one or more processors and configured to train structure and parameterization of each decision tree using training data having a plurality of datasets, each dataset having elements of at least one labeled property, the training executing an objective function that determines the parameters of each decision tree.

A labeling system for generating a trained factor graph, the labeling system comprising:

• processor-readable memory;
• a graph definer configured to define in the processor-readable memory a factor graph including a factor graph model replicated for each variable node of a set of variable nodes, each variable node being informed by one or more factor types in the factor graph model, each factor type being implemented as a single decision tree;
• and
• a graph trainer configured to train structure and parameterization of each decision tree using training data having a plurality of datasets, each dataset having elements of at least one labeled property, the training executing an objective function that determines the parameters of each decision tree, the trained factor graph being stored in the processor-readable memory.