Báo cáo khoa học: "Learning Condensed Feature Representations from Large Unsupervised Data Sets for Supervised Learning" docx

... novel approach for ef- fectively utilizing unsupervised data in addi- tion to supervised data for supervised learn- ing. We use unsupervised data to gener- ate informative condensed feature represen- tations’ ... original features into the condensed features. For this purpose, we define the feature po- tency, which is evaluated by employing an existing supervised m...

Ngày tải lên: 07/03/2014, 22:20

... only the explicitly specified features, but also all available sets of size d of features. For d = 2, this means considering all feature pairs, while for d = 3 all feature triplets. In practice, ... al., 2006). The training data is split into several datasets according to an application specific heuristic. A sep- arate classifier is then trained for each dataset. For example, it...

Ngày tải lên: 17/03/2014, 02:20

... features for the extracted terms and use them to train a classifier that selects the terms that are useful for indexing the associated images. supervised learning scenario with data obtained from ... identifier as a feature because some fre- quently mapped concepts are consistently ineffective for indexing the images in our training and evaluation data. For exam- ple, the CUI...

Ngày tải lên: 17/03/2014, 22:20

... characters, e.g., “鈿鉀鍾” for “鋰”, “裸袖嘿” for “裡”, “湮湩渭＂ for “浬”, “狎猥狠狙” for “狸” , and “黑墨＂ for “里”. We employed similar techniques to recommend characters for In-lists and Out-lists. The database that ... Jhuyin information. In Fig- ure 2, the Jhuyin information is now added beside the sample Chinese words, i.e., “裡面” (li3 mian4). 4 An Open Authoring T...

Ngày tải lên: 19/02/2014, 19:20

... include feature- based and kernel-based classification. Feature- based approaches transform the context of two entities into a liner vector of carefully selected linguistic features, varying from ... entity related information with context information. 3.1 Classification Features The classification is based on the following four types of features. z Entity Positional Structure Fea...

Ngày tải lên: 20/02/2014, 09:20

... the support for eas- ily adding information extracted from further static analysis of the grammar rules, while maintaining the same indexing strategy. Flexibility is one of the reasons for the successful ... used as formal representatives of rich grammatical categories. In this paper, the formal- ism from (Carpenter, 1992) will be used. A TFSG is defined relative to a fixed set of types...

Ngày tải lên: 20/02/2014, 15:21

... only if it does not satisfy ~. For any set • of feature terms, Mod(&) denotes the set of all feature structures for which each E r~ is true. For a formal definition of satisfaction, ... Horn feature clauses A feature literal is ei- ther an atomic feature term (e.g., (~: a), (~ ~ /~), or _L) or its negation. A feature clause is a finite disjunction £lvt~v vl,n...

Ngày tải lên: 22/02/2014, 10:20

... and obtain a condensed set of labeled data points. To this set of labeled data points we add a large number of unlabeled data points labeled by a NN classifier T on the original data set. We use ... labeled and unlabeled data. It finds a condensed set of labeled and unlabeled data points, typically smaller than what is obtained using condensed nearest neighbor on the la- bele...

Ngày tải lên: 07/03/2014, 22:20

... models tend to work better for small datasets and discriminative models tend to work better for larger datasets (Ng and Jordan, 2002). Additionally, they made no use of features, one of the primary benefits ... usually provide much smaller performance gains than the gains possible from good feature engineering. For example, in (Lafferty et al., 2001), when switching from a generati...

Ngày tải lên: 08/03/2014, 01:20

... uniform distribution over all values of y and an empty feature set. For each candidate feature in a predefined feature space, it computes the likelihood gain achieved by includ- ing the feature ... correspond to the selected features. Then, the SGC algorithm is performed on each of the feature groups and new features are se- lected from each of them. In other words, the fe...

Ngày tải lên: 17/03/2014, 04:20