Inspired by previous chunk-level reordering approaches to statistical machine translation, this paper presents two methods to improve the reordering at the chunk level. By introducing a new lattice weighting factor an...
详细信息
Time delays in communication channels usually make teleoperation systems unstable. To cope with this, we apply the energy-bounding algorithm (EBA) that had been proposed for stable haptic interaction. The EBA restrict...
详细信息
Time delays in communication channels usually make teleoperation systems unstable. To cope with this, we apply the energy-bounding algorithm (EBA) that had been proposed for stable haptic interaction. The EBA restricts the energy generated by the sample and hold operator to consumable energy by the energy-consuming physical damping in the haptic system and some part of human arm to guarantee the passivity condition of whole system. This algorithm always guarantees stable haptic interactions, but compromises the displayable impedance range of the virtual environment. The EBA can be straightforwardly applied for bilateral teleoperation due to the analogy between the haptic simulation system and the teleoperation system. The video shows some feasibility test results of the EBA for bilateral teleoperation. Various test results for free, contact, and abrupt motions show that the EBA can ensure stable bilateral teleoperation for the fairy large amount of constant/variable time delays (2.5 sec (one-way) for free motion and 300 msec (one-way) for contact motion).
We present a Minimum Bayes Risk (MBR) decoder for statistical machine translation. The approach aims to minimize the expected loss of translation errors with regard to the BLEU score. We show that MBR decoding on N-be...
详细信息
This paper describes an efficient method to extract large n-best lists from a word graph produced by a statistical machine translation system. The extraction is based on the k shortest paths algorithm which is efficie...
详细信息
In this paper, we describe a source-side reordering method based on syntactic chunks for phrase-based statistical machine translation. First, we shallow parse the source language sentences. Then, reordering rules are ...
详细信息
In phrase-based statistical machine translation, the phrase-table requires a large amount of memory. We will present an efficient representation with two key properties: on-demand loading and a prefix tree structure f...
详细信息
The RWTH system for the IWSLT 2007 evaluation is a combination of several statistical machine translation systems. The combination includes Phrase-Based models, a n-gram translation model and a hierarchical phrase mod...
详细信息
This paper describes the RWTH speech recognition system for Arabic. Several design aspects of the system, including cross-adaptation, multiple system design and combination, are analyzed. We summarize the semi-automat...
详细信息
ISBN:
(纸本)9781424413690;1424413699
This paper describes the RWTH speech recognition system for Arabic. Several design aspects of the system, including cross-adaptation, multiple system design and combination, are analyzed. We summarize the semi-automatic lexicon generation for Arabic using a statistical approach to grapheme-to-phoneme conversion and pronunciation statistics. Furthermore, a novel ASR-based audio segmentation algorithm is presented. Finally, we discuss practical approaches for parallelized acoustic training and memory efficient lattice rescoring. Systematic results are reported on recent GALE evaluation corpora.
This paper describes the development of the RWTH Mandarin LVCSR system. Different acoustic front-ends together with multiple system cross-adaptation are used in a two stage decoding framework. We describe the system i...
详细信息
ISBN:
(纸本)9781424413690;1424413699
This paper describes the development of the RWTH Mandarin LVCSR system. Different acoustic front-ends together with multiple system cross-adaptation are used in a two stage decoding framework. We describe the system in detail and present systematic recognition results. Especially, we compare a variety of approaches for cross-adapting to multiple systems. During the development we did a comparative study on different methods for integrating tone and phoneme posterior features. Furthermore, we apply lattice based consensus decoding and system combination methods. In these methods, the effect of minimizing character instead of word errors is compared. The final system obtains a character error rate of 17.7% on the GALE 2006 evaluation data.
High-throughput comparative genomic hybridization arrays have recently been developed to detect DNA copy number (DCN) aberrations. The DCN data from these arrays is often very noisy and thus calls for appropriate deno...
详细信息
暂无评论