Graph-based transfer learning

摘要

Transfer learning is the task of leveraging the information from labeled examples in some domains to predict the labels for examples in another domain. It finds abundant practical applications, such as sentiment prediction, image classification and network intrusion detection. A graph-based transfer learning framework propagates label information from a source domain to a target domain via the example-feature-example tripartite graph, and puts more emphasis on the labeled examples from the target domain via the example-example bipartite graph. An iterative algorithm renders the framework scalable to large-scale applications. The framework propagates the label information to both features irrelevant to the source domain and unlabeled examples in the target domain via common features in a principled way.

主权项

1. A computer method, comprising carrying out operations on a computer, the operations comprising:
maintaining machine readable embodiments on a medium of a bipartite graph and a tripartite graph,
the tripartite graph comprising
a first plurality of nodes corresponding to labeled and unlabeled examples from source and target domains;a second plurality of nodes corresponding to features; anda first plurality of edges connecting
the nodes corresponding to the featuresto the nodes corresponding to the examplesaccording to whether the features appear in the examples or not;the bipartite graph comprising
the first plurality of nodes corresponding to the examples; anda second plurality of edges connecting the examples, the edges being associated with indications that indicate whether connected examples are in a same domain or not; deriving labels for at least one target domain based on the tripartite and bipartite graphs; and presenting an embodiment of the labels as a result, wherein said deriving comprises:
formulating an objective function based on said bipartite and tripartite graphs, said objective function encompassing smoothness and consistency constraints and providing label information in the target domain at least responsive to label information in the source domain;
applying the objective function to the all examples, whether labeled or unlabeled, and all features in order to obtain at least one result relative to the unlabeled examples;minimizing the objective function to yield a label function; andproviding output labels responsive to the label function.