Recent publications:

• Computational methods are invaluable for typology, but the models must match the questions: Commentary on Dunn et al. (2011) [Levy+HD] (2011)@Misc{daume11dunn,
     author = {Roger Levy and Hal {Daum\'e III}},
     title = {Computational methods are invaluable for typology, but the models must match the questions: Commentary on Dunn et al. (2011)},
     howpublished = {Journal of Linguistic Typology},
     year = {2011},
     url = {http://hal3.name/docs/#daume11dunn}
  }
  
• Active Supervised Domain AdaptationAbstract     In this paper, we harness the synergy between two important learning paradigms, namely, active learning and domain adaptation. We show how active learning in a target domain can leverage information from a different but related source domain. Our proposed framework, Active Learning Domain Adapted (ALDA), uses source domain knowledge to transfer information that facilitates active learning in the target domain. We propose two variants of ALDA: a batch B-ALDA and an online O-ALDA. Empirical comparisons with numerous baselines on real-world datasets establish the efficacy of the proposed methods. [Saha+al.] (ECML 2011)@InProceedings{daume11alda,
     author = {Avishek Saha and Piyush Rai and Hal {Daum\'e III} and Suresh Venkatasubramanian and Scott L. DuVall},
     title = {Active Supervised Domain Adaptation},
     booktitle = {European Conference on Machine Learning (ECML)},
     year = {2011},
     address = {Athens, Greece},
     tags = {ml da},
     url = {http://hal3.name/docs/#daume11alda},
  }
     
  
• Corpus-Guided Sentence Generation of Natural ImagesAbstract     We propose a sentence generation strategy that describes images by predicting the most likely nouns, verbs, scenes and prepositions that make up the core sentence structure. The input are initial noisy estimates of the objects and scenes detected in the image using state of the art trained detectors. As predicting actions from still images directly is unreliable, we use a language model trained from the English Gigaword corpus to obtain their estimates; together with probabilities of co-located nouns, scenes and prepositions. We use these estimates as parameters on a HMM that models the sentence generation process, with hidden nodes as sentence components and image detections as the emissions. Experimental results show that our strategy of combining vision and language produces readable and descriptive sentences compared to naive strategies that use vision alone. [Yang+al.] (EMNLP 2011)@InProceedings{daume11generation,
     author = {Yezhou Yang and Ching Lik Teo and Hal {Daum\'e III} and Yiannis Aloimonos},
     title = {Corpus-Guided Sentence Generation of Natural Images},
     booktitle = {Empirical Methods in Natural Language Processing (EMNLP)},
     year = {2011},
     address = {Edinburgh, Scotland},
  }
  
• Beam Search based MAP Estimates for the Indian Buffet Process [Rai+HD] (ICML 2011)@InProceedings{daume11ibpsearch,
     author = {Piyush Rai and Hal {Daum\'e III}},
     title = {Beam Search based MAP Estimates for the Indian Buffet Process},
     booktitle = {International Conference on Machine Learning (ICML)},
     year = {2011},
     address = {Bellevue, WA},
     url = {http://hal3.name/docs/#daume11ibpsearch}
  }
  
• A Co-training Approach for Multiview Spectral Clustering [Kumar+HD] (ICML 2011)@InProceedings{daume11cospec,
     author = {Abhishek Kumar and Hal {Daum\'e III}},
     title = {A Co-training Approach for Multiview Spectral Clustering},
     booktitle = {International Conference on Machine Learning (ICML)},
     year = {2011},
     address = {Bellevue, WA},
     url = {http://hal3.name/docs/#daume11cospec}
  }
  
• Improving Bilingual Projections via Sparse Covariance Matrices [Jagarlamudi+al.] (EMNLP 2011)@InProceedings{daume11sparse,
     author = {Jagadeesh Jagarlamudi and Raghavendra Udupa and Hal {Daum\'e III} and Abhijit Bhole},
     title = {Improving Bilingual Projections via Sparse Covariance Matrices},
     booktitle = {Empirical Methods in Natural Language Processing (EMNLP)},
     year = {2011},
     address = {Edinburgh, Scotland},
  }
  
• Approximate Scalable Bounded Space Sketch for Large Data NLP [Goyal+HD] (EMNLP 2011)@InProceedings{daume11sketch,
     author = {Amit Goyal and Hal {Daum\'e III}},
     title = {Approximate Scalable Bounded Space Sketch for Large Data {NLP}},
     booktitle = {Empirical Methods in Natural Language Processing (EMNLP)},
     year = {2011},
     address = {Edinburgh, Scotland},
     url = {http://hal3.name/docs/#daume11sketch}
  }
  
• Lossy Conservative Update (LCU) sketch: Succinct approximate count storage [Goyal+HD] (AAAI 2011)@InProceedings{daume11lcu,
     author = {Amit Goyal and Hal {Daum\'e III}},
     title = {Lossy Conservative Update ({LCU}) sketch: Succinct approximate count storage},
     booktitle = {Conference on Artificial Intelligence (AAAI)},
     year = {2011},
     address = {Portland, OR},
     url = {http://hal3.name/docs/#daume11lcu}
  }
  
• Online Learning of Multiple Tasks and Their RelationshipsAbstract     We propose an Online MultiTask Learning (OMTL) framework which simultaneously learns the task weight vectors as well as the task relatedness adaptively from the data. Our work is in contrast with prior work on online multitask learning which assumes fixed task relatedness, a priori. Furthermore, whereas prior work in such settings assume only positively correlated tasks, our framework can capture negative correlations as well. Our proposed framework learns the task relationship matrix by framing the objective function as a Bregman divergence minimization problem for positive definite matrices. Subsequently, we exploit this adaptively learned task-relationship matrix to select the most informative samples in an online multitask active learning setting. Experimental results on a number of real-world datasets and comparisons with numerous baselines establish the efficacy of our proposed approach. [Saha+al.] (AI-Stats 2011)@InProceedings{daume11olmt,
     author = {Avishek Saha and Piyush Rai and Hal {Daum\'e III} and Suresh Venkatasubramanian},
     title = {Online Learning of Multiple Tasks and Their Relationships},
     booktitle = {Conference on Artificial Intelligence and Statistics (AI-Stats)},
     year = {2011},
     address = {Ft. Lauderdale, FL},
     url = {http://hal3.name/docs/#daume11olmt}
  }
  
• Domain Adaptation for Machine Translation by Mining Unseen WordsAbstract     We show that unseen words account for a large part of the translation error when moving to new domains. Using an extension of a recent approach to mining translations from comparable corpora (Haghighi et al., 2008), we are able to find translations for otherwise OOV terms. We show several approaches to integrating such translations into a phrasebased translation system, yielding consistent improvements in translations quality (between 0.5 and 1.5 Bleu points) on four domains and two language pairs. [HD+Jagarlamudi] (2011)@InProceedings{daume11lexicaladapt,
     author = {Hal {Daum\'e III} and Jagadeesh Jagarlamudi},
     title = {Domain Adaptation for Machine Translation by Mining Unseen Words},
     booktitle = {Association for Computational Linguistics},
     year = {2011},
     address = {Portland, OR},
     url = {http://hal3.name/docs/#daume11lexicaladapt}
  }