Báo cáo khoa học: "Semi-supervised Learning for Automatic Prosodic Event Detection Using Co-training Algorithm" doc

... co-training algorithm to automatic prosodic event detection and propose methods to better select samples to improve semi- supervised learning performance for this task. 4 Prosodic event detection method We ... 540–548, Suntec, Singapore, 2-7 August 2009. c 2009 ACL and AFNLP Semi-supervised Learning for Automatic Prosodic Event Detection Using Co-training...

Ngày tải lên: 23/03/2014, 16:21

... our learning algorithm is the strategy used to select samples for training. In general, this novel learning framework lies be- tween supervised learning and reinforcement learn- ing. Guided learning ... classification are dynamically incorporated in the learning phase. Guided learning is not as hard as reinforcement learning. At each local step in learning, we always know the u...

Ngày tải lên: 20/02/2014, 12:20

... rely on additional labeled data for learning the ranker; instead, we use half of the training texts for training classifiers and the other half for ranking purposes. Results using our approach are shown ... and Cardie (2002b) for details. Learning algorithms. We consider three learning algorithms, namely, the C4.5 decision tree induction system (Quinlan, 1993), the RIPPER rule...

Ngày tải lên: 20/02/2014, 15:20

... per- formance to systems using sophisticated parse trees. 1 Introduction Function tags, such as subject, object, time, loca- tion, etc. are conceptually appealing by encoding an event in the format ... applying it to our specific task. 3.1 Learning Model The framework from which HM-SVM are derived is a maximum margin formulation for joint fea- ture functions in kernel learning settin...

Ngày tải lên: 08/03/2014, 01:20

... N- best list. For details, see (Ueffing and Ney, 2007). Again, the single values are multiplied to obtain a score for the whole sentence. For NIST, the lan- guage model score is determined using a 5-gram model ... 24.4 for the 100K filtered corpus, and from 24.5 to 24.8 for 150K, respectively. Moreover, we see that the BLEU score of the system using 100K training sen- tence pairs...

Ngày tải lên: 08/03/2014, 02:21

... paper we study spectral learning methods for non-deterministic split head- automata grammars, a powerful hidden- state formalism for dependency parsing. We present a learning algorithm that, ... parsing and learning algorithms for grammatical bigram models. Techni- cal Report UCB/CSD-01-1148, University of Cali- fornia, Berkeley. Slav Petrov and Dan Klein. 2007. Improved infer- en...

Ngày tải lên: 08/03/2014, 21:20

... 128 documents, divided into 70 for training, 18 for development, and 40 for test. In the puzzle game domain, we use 50 tutorials, divided into 40 for training and 10 for test. 9 Statistics for ... contrast, our emphasis is on learning language by proactively interacting with an external environment. Reinforcement Learning for Language Pro- cessing Reinforcement learning ha...

Ngày tải lên: 17/03/2014, 01:20

... a better learning curve. 4.4 Summary Result Figure 8 shows the best active learning result compared with that of random selection. The learning curve for ac- tive learning is obtained using as ... to achieve a satisfactory performance using active learning. Active learning has been studied in the context of many natural language processing (NLP) applications such as informat...

Ngày tải lên: 17/03/2014, 08:20

Ngày tải lên: 23/03/2014, 14:20

... shows the performance of random sentence selection for AL combined with self-training /co-training for the multi-source translation from the four Germanic languages to English. It shows that the co-training mode ... Work (Haffari et al., 2009) provides results for active learning for MT using a single language pair. Our work generalizes to the use of multilingual corpora using...

Ngày tải lên: 23/03/2014, 16:21