Tài liệu Báo cáo khoa học: "Learning to Translate with Multiple Objectives" doc

... LinguisticsLearning to Translate with Multiple ObjectivesKevin Duh∗Katsuhito Sudoh Xianchao Wu Hajime Tsukada Masaaki NagataNTT Communication Science Laboratories2-4 Hikari-dai, Seika-cho, Kyoto 619-0237, ... between pareto vs. non-paretopoints ignores the fact that 2nd-place non-paretopoints may also lead to good practical solutions. Abetter approach may be to adopt a graded deﬁnitionof Pareto optimality ... deﬁnitely lead to new insights. (3) Finally, it would be interesting to explore other creative uses of multiple- objectives inMT beyond multiple metrics. For example: Can welearn to translate faster...

Tài liệu Báo cáo khoa học: "Learning to Find Translations and Transliterations on the Web" doc

... Emmy Awards) to query a search engine with the target webpage language set to the target language (e.g., Chinese), biasing the search engine to return Chinese webpages interspersed with some ... information to generate transliteration feature for a Chinese token with respect to English words in the query. We extract person or location entries in Wikipedia as name transliterated pairs to generate ... the data labeled with translation tags and three kinds feature values to train a CRF model. 3.2 Run-Time Translation Extraction With the trained CRF model, we then attempt to find translations...

Tài liệu Báo cáo khoa học: "Learning to Follow Navigational Directions" pdf

... itis her task to describe the path to the instructionfollower, who cannot see the reference path. Oursystem learns to interpret these navigational direc-tions, without access to explicit linguistic ... Compare“go to the uh right of the mine”which utilizes an allocentric frame, with “right then go eh uh to your right hori-zontally”which uses an egocentric frame of reference. Itis difﬁcult to distinguish ... Re-search Fellowship to the ﬁrst author and awardIIS-0811974 to the second author and by the AirForce Research Laboratory (AFRL), under primecontract no. FA8750-09-C-0181. Thanks to Michael Levit...

Tài liệu Báo cáo khoa học: "Learning to Compose Effective Strategies from a Library of Dialogue Components" doc

... is to listen to the user. This paper de-scribes the development of an adaptive dialogue sys-tem that uses the feedback of users to automaticallyimprove its strategy. The system starts with ... possible (but with enough distinc-793alogue policy, experts can impossibly take into ac-count all possible different dialogue contexts. With respect to future work, we plan to examinethe ... been limited to welldeﬁned areas of the dialogue, in particular dealing with speech recognition and clariﬁcation problems, with small state spaces and a limited set of actions to choose from...

Tài liệu Báo cáo khoa học: "Learning to Recognize Tables in Free Text" pptx

... separator characters. F9 The number of segments in H with three or more contiguous separator characters. Table 1: Feature values for table boundary space (or special) characters to a vline with ... 12: 13: Net tons Capability 14: produced utilization 15: Week to March 14 1,633,000 75.8~ 16: Week to March 7 1,570,000 71.9~ 17: Year to date 15,029,000 66.9~ 18: Year earlier to date 18,431,000 ... original documents, we man- ually annotated the plain-text documents with table boundary, column, and row information. The documents shown in Figure 1 and 2 are part of the 100 documents...

Tài liệu Báo cáo khoa học: "Learning to Behave by Reading" pot

...

Tài liệu Báo cáo khoa học: "Learning Event Durations from Event Descriptions" docx

... n tokens to its left and n to- kens to its right, as well as the event itself, for n = {0, 1, 2, 3}. The best n determined via cross validation turned out to be 0, i.e., the event itself with ... from 5 to 7 in the natural logarithmic scale, which corresponds to about 1.5 minutes to 30 minutes. The other is from 14 to 17 in the natural logarithmic scale, which corresponds to about ... total annotated non-WSJ data (2132 event instances) into two data sets: a train-ing data set with 1705 event instances (about 80% of the total non-WSJ data) and a held-out test data set with...

Tài liệu Báo cáo khoa học: "Learning Word Senses With Feature Selection and Order Identiﬁcation Capabilities" pdf

... contextual word’s vectorstogether may result in noise-dominated second or-der context vectors. To deal with this problem, we extend the featureselection procedure further to the construction ... was conducted to reduce the di-mensionality of context vectors. Then the reducedcontext vectors were grouped into a pre-definednumber of clusters whose centroids corresponded to senses of target ... goodness of fit ofa model to data sample, while the second term pe-nalizes complex model. This estimator works by at-tempting to find a model order with minimum codelength to describe the data sample...

Tài liệu Báo cáo khoa học: "TEMPORALON TO LOGY IN NATURAL LANGUAGE" doc

... temporal/aspectual adverbials appear to do is to transform entities of one type into these other "contingently" related entities, or to turn them into composites with those related entities. ... conjunction with some coercive aspectuzl category which forces it to undergo a transition, then the alter- natives that are available are: a) to decompose the core event into a nucleus and to make ... transition to one of the componants, such as the prepuratory activity of climbing or to the consequent state of having climbed the mountain; or b) to treat the entire event as a culmination, to c(m~oose...

Tài liệu Báo cáo khoa học: "K-means Clustering with Feature Hashing" docx

... equivalently Eφ[||x||2φ−||x||2] = 0.In order to show distortion, we want to use Cheby-shev’s inequality. To this end, it is vital to know theexpectation and variance of the sum of squared ... are dense vectors where Kis the number of clusters. Thus, it is infeasible to store them in memory and slow to compute if the di-mension of inputs is huge, as is often the case with NLP and ... K-means is a process which monoton-ically decreases RSS in each step, if RSS is not dis-torted so much by feature hashing, we can expectresults to be reliable to some extent.Let us deﬁne the...

Xem thêm