Loading...
Unsupervised domain adaptation via representation learning and adaptive classifier learning
Gheisari, M ; Sharif University of Technology
903
Viewed
- Type of Document: Article
- DOI: 10.1016/j.neucom.2015.03.020
- Abstract:
- The existing learning methods usually assume that training data and test data follow the same distribution, while this is not always true. Thus, in many cases the performance of these methods on the test data will be severely degraded. In this paper, we study the problem of unsupervised domain adaptation, where no labeled data in the target domain is available. The proposed method first finds a new representation for both the source and the target domain and then learns a prediction function for the classifier by optimizing an objective function which simultaneously tries to minimize the loss function on the source domain while also maximizes the consistency of manifold (which is based on the nature of data) with the prediction function. We run the proposed method on four types of cross-domain image classification problems and show that the proposed method significantly outperforms several state-of-the-art domain adaptation methods
- Keywords:
- Neural networks ; Adaptive classifiers ; Domain adaptation ; Learning methods ; Objective functions ; Prediction function ; Representation learning ; Source domain ; Target domain ; Image classification ; Adaptive classification and representation learning ; Article ; Classification algorithm ; Controlled study ; Domain adaptation algorithm ; Geodesic flow kernel ; Grassmannian manifold ; Learning algorithm ; Machine learning ; Prediction ; Principal component analysis ; Priority journal ; Statistical parameters
- Source: Neurocomputing ; Volume 165 , 2015 , Pages 300-311 ; 09252312 (ISSN)
- URL: http://www.sciencedirect.com/science/article/pii/S0925231215002921