Fast and Scalable Training of Semi-Supervised CRFswith Application to Activity Recognition.pdfVIP

Fast and Scalable Training of Semi-Supervised CRFs with Application to Activity Recognition Maryam Mahdaviani Computer Science Department University of British Columbia Vancouver, BC Canada Tanzeem Choudhury Intel Research 1100 NE 45th Street Seattle, WA 98105 USA Abstract We present a new and efficient semi-supervised training method for parameter es- timation and feature selection in conditional random fields (CRFs). In real-world applications such as activity recognition, unlabeled sensor traces are relatively easy to obtain whereas labeled examples are expensive and tedious to collect. Furthermore, the ability to automatically select a small subset of discriminatory features from a large pool can be advantageous in terms of computational speed as well as accuracy. In this paper, we introduce the semi-supervised virtual evidence boosting (sVEB) algorithm for training CRFs – a semi-supervised extension to the recently developed virtual evidence boosting (VEB) method for feature selection and parameter learning. The objective function of sVEB combines the unlabeled conditional entropy with labeled conditional pseudo-likelihood. It reduces the overall system cost as well as the human labeling cost required during training, which are both important considerations in building real-world inference systems. Experiments on synthetic data and real activity traces collected from wearable sensors, illustrate that sVEB benefits from both the use of unlabeled data and au- tomatic feature selection, and outperforms other semi-supervised approaches. 1 Introduction Conditional random fields (CRFs) are undirected graphical models that have been successfully ap- plied to the classification of relational and temporal data [1]. Training complex CRF models with large numbers of input features is slow, and exact inference is often intractable. The ability to select the most informative features as needed can reduce the training time and the risk of over-fitting of parameters. Furtherm