Transfer learning using task-level features with application to information retrieval

摘要: We propose a probabilistic transfer learning model that uses task-level features to control the task mixture selection in hierarchical Bayesian model. These features, although rarely used existing approaches, can provide additional information complex distributions and allow effective new tasks especially when only limited number of data are available. To estimate parameters, we develop an empirical Bayes method based on variational approximation techniques. Our experiments retrieval show proposed achieves significantly better performance compared with other methods.