Báo cáo khoa học: "Weakly Supervised Learning for Hedge Classification in Scientific Literature" pot

... weakly supervised ML, and point toward avenues for future research. 1 Introduction The automatic processing of scientific papers using NLP and machine learning (ML) techniques is an increasingly ... addressed in section 10.3. In this study we develop a learning model based around the concept of iteratively predicting labels for unlabelled training samples, the basic paradigm f...

Ngày tải lên: 23/03/2014, 18:20

... No_Sharing _for_ Age and Conflict _for_ Affiliation] }. The induced features provide direct and fine- grained information, but suffer from less sampling space. Combining basic features and induced ... discretized into integers. The number of integers being used may impact the final performance of the system. If the number is too small, significant information may be lost during the...

Ngày tải lên: 31/03/2014, 03:20

... two phases: 1. Training the Classifiers We independently train binary classifiers for each semantic re- lation using both shallow and deep features. 2. Ensemble Learning and Ranking Each of the five ... classifiers are combined using ensem- ble learning techniques to determine the most confident classification. The obtained list of the classified instances is ranked using pat- tern scores, i...

Ngày tải lên: 07/03/2014, 22:20

Ngày tải lên: 08/03/2014, 01:20

... significant performance gain according to the size of initial labeled training data. Finally we compared our co-training perfor- mance with supervised learning. For supervised learning, all labeled ... for the test set are used for training. We used most confi- dent selection with proposed self-labeling method. The initial training data in co-training is 3% of that used for sup...

Ngày tải lên: 23/03/2014, 16:21

... appear in Corpus (in our experimen- tal settings - at least twice). For example, for the class “mountain” training examples are: “Ever- est”, “Mauna Loa”, etc. The algorithm learns from each training ... divided in two classes, according to different use of train- ing data: Unsupervised approaches (see (Cimi- ano et al., 2005) for details) and supervised ap- proaches which use ma...

Ngày tải lên: 31/03/2014, 20:20

... domains. For example, the training method used in SPoT (Sentence Planner Train- able), as described in (Walker, Rambow and Ro- gati, 2002), was only shown to work in the travel domain, for the information ... trainable approach is feasible by showing (1) that the training tech- nique used for SPoT can be extended to a new domain (restaurant information); (2) that this technique, p...

Ngày tải lên: 23/03/2014, 19:20

... histi- dine involved in covalent FAD binding, but contains a leucine instead (Leu56). Leu56 replacements did not result in covalent flavinylation but revealed the importance of Leu56 for both FAD-binding ... targeting sequence in plant GALDH (FR ⁄ YA). The asterisk (*) marks the histidine residue involved in covalent binding of the FAD cofactor in GUO, ALO and GLO. The FAD-binding...

Ngày tải lên: 23/03/2014, 07:20

... created by perturbing e). Wang and Schuurmans (2005) improve an unsupervised HMM-based tagger by constraining the learned structure to maintain appropriate marginal tag probabilities and using word similarities ... Bayesian inference for POS tagging. However, their work departs from exist- ing Bayesian approaches to POS tagging in that they (1) introduce a new sparse prior on the dis- tri...

Ngày tải lên: 24/03/2014, 03:20

... These indirect methods for leveraging prior knowledge can be cumbersome and unintuitive for a non-machine -learning expert. This paper proposes a method for directly guid- ing the learning of ... constraints accuracy Spanish (maximum length 60) constraint baseline GE Figure 3: Comparing GE training of a CRF and constraint baseline while increasing the number of oracle constraints...

Ngày tải lên: 20/02/2014, 07:20