Word segmentation is helpful in Chinese natural language processing in many aspects. However it is showed that different word segmentation strategies do not affect the performance of Statistical Machine Translation (S...
详细信息
Word segmentation is helpful in Chinese natural language processing in many aspects. However it is showed that different word segmentation strategies do not affect the performance of Statistical Machine Translation (SMT) from English to Chinese significantly. In addition, it will cause some confusions in the evaluation of English to Chinese SMT. So we make an empirical attempt to translation English to Chinese in the character level, in both the alignment model and language model. A series of empirical comparison experiments have been conducted to show how different factors affect the performance of character-level English to Chinese SMT. We also apply the recent popular continuous s- pace language model into English to Chinese SMT. The best performance is obtained with the BLEU score 41.56, which improve base- line system (40.31) by around 1.2 BLEU s- core.
Neural network language models (NNLMs) have been shown to outperform traditional n-gram language model. However, too high computational cost of NNLMs becomes the main obstacle of directly integrating it into pinyin IM...
详细信息
Neural network language models (NNLMs) have been shown to outperform traditional n-gram language model. However, too high computational cost of NNLMs becomes the main obstacle of directly integrating it into pinyin IME that normally requires a real-Time response. In this paper, an efficient solution is proposed by converting NNLMs into back-off n-gram language models, and we integrate the converted NNLM into pinyin IME. Our exper-imental results show that the proposed method gives better decoding predictive performance for pinyin IME with satisfied efficiency.
In this work, we present a novel way of using neural network for graph-based dependency parsing, which fits the neural network into a simple probabilistic model and can be furthermore generalized to high-order parsing...
详细信息
In this work, we present a novel way of using neural network for graph-based dependency parsing, which fits the neural network into a simple probabilistic model and can be furthermore generalized to high-order parsing. Instead of the sparse features used in traditional methods, we utilize distributed dense feature representations for neural network, which give better feature representations. The proposed parsers are evaluated on English and Chinese Penn Treebanks. Compared to existing work, our parsers give competitive performance with much more efficient inference.
作者：
Yingxu WangLaboratory for Computational Intelligence and Software Science
International Institute of Cognitive Informatics and Cognitive Computing (ICIC) Department of Electrical and Computer Engineering Schulich School of Engineering and Hotchkiss Brain Institute University of Calgary Calgary Canada
In this paper, we propose a scalable clustering paradigm to address the problems of excessive computational load and limited clustering performance in large-scale data. The proposed method employs the enhanced splitti...
详细信息
We investigate the noise reduction performance for event related optical signal (EROS) measured from 5-channel near-infrared spectroscopy to reduce noise by using grand average and independent component analysis (ICA)...
详细信息
We investigate the noise reduction performance for event related optical signal (EROS) measured from 5-channel near-infrared spectroscopy to reduce noise by using grand average and independent component analysis (ICA)-based Kalman filter. Episodic memory retrieval task was performed and expected to show brain activity around 500 ms from right prefrontal cortex. We verified that positive deviation in response to old word occurred around 500 ms using grand average for three subjects. After applying ICA-based Kalman filter on data from each subject, the clear positive deviation was shown graphically. Also, the ratio of positively deviated epoch in response to old word increased by 14 % after ICA-based Kalman filtering.
We introduce a unifying generalization of the Lovasz theta function, and the associated geometric embedding, for graphs with weights on both nodes and edges. We show how it can be computed exactly by semidefinite prog...
详细信息
ISBN:
(纸本)9781510825024
We introduce a unifying generalization of the Lovasz theta function, and the associated geometric embedding, for graphs with weights on both nodes and edges. We show how it can be computed exactly by semidefinite programming, and how to approximate it using SVM computations. We show how the theta function can be interpreted as a measure of diversity in graphs and use this idea, and the graph embedding in algorithms for Max-Cut, correlation clustering and document summarization, all of which are well represented as problems on weighted graphs.
In the new era of intelligence revolution beyond information and computing revolutions,a brain-like intelligent system is emerging known as cognitive robots.A cognitive robot is an autonomous system that is capable of...
详细信息
In the new era of intelligence revolution beyond information and computing revolutions,a brain-like intelligent system is emerging known as cognitive robots.A cognitive robot is an autonomous system that is capable of perception,learning,and inference mimicking the cognitive mechanisms of the brain by computational *** robotics is a discipline that studies theories,mathematical means,implementations,and engineering applications of cognitive *** theoretical foundations of cognitive robotics are underpinned in brainscience,cognitivescience,abstract intelligence,knowledge science,intelligence science,cognitive linguistics,and denotational *** keynote lecture presents the latest advances in theories and applications of cognitive robotics and brain-inspired *** theoretical framework of cognitive robotics and its denotational mathematical means such as concept algebra,semantic algebra,behavioral process algebra,and inference algebra are *** challenges in cognitive robotics are formally studied on what the necessary and sufficient intelligent behaviors of cognitive robots are,and what distinguish the intelligent capabilities of cognitive robots from their imperative *** structural and functional models of cognitive robots are explored at the intelligent layers of imperative,autonomic,and cognitive *** progress in cognitive robotics has led to the development of cognitive systems,nextgeneration computers,and cognitive machine learning engines,as well as the growth of contemporary knowledge-based industries.
How do we perceive the predictability of functions? We derive a rational measure of a function's predictability based on Gaussian process learning curves. Using this measure, we show that the smoothness of a funct...
详细信息
brain activity is usually measured by non-invasive modalities. Inter alia, the electroencephalogram (EEG) is used most commonly. However, EEG is very sensitive to other biosignals, so other bio-signal detection modali...
详细信息
ISBN:
(纸本)9781479974962
brain activity is usually measured by non-invasive modalities. Inter alia, the electroencephalogram (EEG) is used most commonly. However, EEG is very sensitive to other biosignals, so other bio-signal detection modalities must be used as supplementary systems. Functional near-infrared spectroscopy (fNIRS) has good characteristics for use as such a supplementary modality, because brain activities can be measured by fNIRS through hemodynamic responses. Therefore, many scientists have adopted fNIRS for brain machine interface (BCI). Recently, fNIRS has become more compact and is robust to noise, so it could bring us to the development of an effective wearable BCI.
暂无评论