Domain Adaptation Workshop: Theory and Application at NIPS 2011 Invited Talk: Discrepancy and Adaptation by Mehryar Mohri Mehryar Mohri is a Professor at the Courant Institute and a Research Consultant at Google. His current research interests include machine learning, computational biology, and text and speech processing.

Domain Adaptation Workshop: Theory and Application at NIPS 2011 Invited Talk: Training Structured Prediction Models with Extrinsic Loss Functions by Slav Petrov Slav Petrov is a Research Scientist at Google New York who works on problems at the intersection of natural language processing and machine learning. In particular, he's interested in syntactic parsing and its applications to machine translation and information extraction. Abstract: We present an online learning algorithm for training structured prediction models with extrinsic loss functions. This allows us to extend a standard supervised learning objective with additional loss-functions, either based on intrinsic or task-specific extrinsic measures of quality. We present experiments with sequence models on part-of-speech tagging and named entity recognition tasks, and with syntactic parsers on dependency parsing and machine translation reordering tasks.

Domain Adaptation Workshop: Theory and Application at NIPS 2011 Invited Speaker: Adaptation without Retraining by Dan Roth Dan Roth is a Professor in the Department of Computer Science and the Beckman Institute at the University of Illinois at Urbana-Champaign and a University of Illinois Scholar. He is also a Fellow of AAAI, for his contributions to the foundations of machine learning and inference and for developing learning centered solutions for natural language processing problems. Abstract: Natural language models trained on labeled data from one domain do not perform well on other domains. Most adaptation algorithms proposed in the literature train a new model for the target domain using a mix of labeled and unlabeled data. We discuss some limitations of existing general purpose adaptation algorithms that are due to the interaction betweendifferences in base feature statistics and task differences and illustrate howthis should be taken into account jointly. With these insights we propose a new approach to adaptation that avoids the need for retraining models. I nstead, at evaluation time, we perturb the given instance to be more similar to instances the model can h andle well, or perturb the model outcomes to fit our expectation of the target domain better, given some prior knowledge on the task and the target domain. We provide experimental evidence in a range of natural language processing, including semantic role labeling and English as a Second Language (ESL ...

Domain Adaptation Workshop: Theory and Application at NIPS 2011 Invited Speaker: On the utility of unlabeled samples in Domain Adaptation by Shai Ben-David Shai Ben-David grew up in Jerusalem, Israel and attended the Hebrew University studying physics, mathematics and psychology. He received his PhD under the supervision of Saharon Shelah and Menachem Magidor for a thesis in set theory (on non-provability of infinite combinatorial statements). In August 2004 he joined the School of Computer Science at the University of Waterloo. Abstract: In many domain adaptation applications, on top of a sample of labeled points generated by the training tasks, the learner can also access unlabeled samples generated by the target distribution. The focus of this talk is to investigate when can such unlabeled samples be (provably) beneficial to the leaner. We show that depending on the type of prior knowledge available to the leaner, there are setups in which unlabeled target-generated samples can make a big difference in the required size of labeled training samples, while in other scenarios such unlabeled samples do not improve the learning rate.

Domain Adaptation Workshop: Theory and Application at NIPS 2011 Invited Talk: History Dependent Domain Adaptation by Allen Lavoie Abstract: We study a novel variant of the domain adaptation problem, in which the loss function on test data changes due to dependencies on prior predictions. One important instance of this problem area occurs in settings where it is more costly to make a new error than to repeat a previous error. We propose several methods for learning effectively in this setting, and test them empirically on the real-world tasks of malicious URL classification and adversarial advertisement detection.