This paper presents a word alignment circuit for high speed SerDes system.By using pipeline structure and circuit optimization techniques,the speed of the aligner is increased,and its performance is improved further t...This paper presents a word alignment circuit for high speed SerDes system.By using pipeline structure and circuit optimization techniques,the speed of the aligner is increased,and its performance is improved further through adopting the full custom design method.The proposed word aligner has fabricated in 0.18μm CMOS technology with total area of 1.075 ×0.775mm^2￣ including I/O pad.Measurement results show that this circuit achieves the maximum data rate of 14.5Gb/s,while consuming a total power of 34.9mW from a 1.8V supply.展开更多
Sentence alignment is a basic task in natural lan-guage processing which aims to extract high-quality paral-lel sentences automatically.Motivated by the observation that aligned sentence pairs contain a larger number ...Sentence alignment is a basic task in natural lan-guage processing which aims to extract high-quality paral-lel sentences automatically.Motivated by the observation that aligned sentence pairs contain a larger number of aligned words than unaligned ones,we treat word translation as one of the most useful external knowledge.In this paper,we show how to explicitly integrate word translation into neural sentence alignment.Specifically,this paper proposes three cross-lingual encoders to incorporate word translation:1)Mixed Encoder that learns words and their translation annotation vectors over sequences where words and their translations are mixed alterma-tively;2)Factored Encoder that views word translations as fea-tures and encodes words and their translations by concatenating their embeddings;and 3)Gated Encoder that uses gate mechanism to selectively control the amount of word translations moving forward.Experimentation on NIST MT and Opensub-titles Chinese-English datasets on both non-monotonicity and monotonicity scenarios demonstrates that all the proposed encoders significantly improve sentence alignment performance.展开更多
This paper claries the definition of alignment from the viewpoint of linguistic similarity. Many alignment algorithms have been proposed with very high precision. But the languages belong to occidental family. We prop...This paper claries the definition of alignment from the viewpoint of linguistic similarity. Many alignment algorithms have been proposed with very high precision. But the languages belong to occidental family. We propose a new method for alignment between languages that do not belong to the same language family. On the contrary to most of the previously proposed methods that rely heavily on statistics, our method attempts to use linguistic knowledge to overcome the problems of statistical model. Experimental results confirm that the algorithm can align over 85 % of word pairs while maintaining a comparably high precision rate, even when a small corpus is used in training.展开更多
基金Supported by the National High Technology Research and Development Programme of China(No.2011AA10305)
文摘This paper presents a word alignment circuit for high speed SerDes system.By using pipeline structure and circuit optimization techniques,the speed of the aligner is increased,and its performance is improved further through adopting the full custom design method.The proposed word aligner has fabricated in 0.18μm CMOS technology with total area of 1.075 ×0.775mm^2￣ including I/O pad.Measurement results show that this circuit achieves the maximum data rate of 14.5Gb/s,while consuming a total power of 34.9mW from a 1.8V supply.
基金This work was supported by the National Natural Science Foundation of China(Grant Nos.61876120,61673290).
文摘Sentence alignment is a basic task in natural lan-guage processing which aims to extract high-quality paral-lel sentences automatically.Motivated by the observation that aligned sentence pairs contain a larger number of aligned words than unaligned ones,we treat word translation as one of the most useful external knowledge.In this paper,we show how to explicitly integrate word translation into neural sentence alignment.Specifically,this paper proposes three cross-lingual encoders to incorporate word translation:1)Mixed Encoder that learns words and their translation annotation vectors over sequences where words and their translations are mixed alterma-tively;2)Factored Encoder that views word translations as fea-tures and encodes words and their translations by concatenating their embeddings;and 3)Gated Encoder that uses gate mechanism to selectively control the amount of word translations moving forward.Experimentation on NIST MT and Opensub-titles Chinese-English datasets on both non-monotonicity and monotonicity scenarios demonstrates that all the proposed encoders significantly improve sentence alignment performance.
文摘This paper claries the definition of alignment from the viewpoint of linguistic similarity. Many alignment algorithms have been proposed with very high precision. But the languages belong to occidental family. We propose a new method for alignment between languages that do not belong to the same language family. On the contrary to most of the previously proposed methods that rely heavily on statistics, our method attempts to use linguistic knowledge to overcome the problems of statistical model. Experimental results confirm that the algorithm can align over 85 % of word pairs while maintaining a comparably high precision rate, even when a small corpus is used in training.
