36 resultados para technology adaptation
em Cambridge University Engineering Department Publications Database
Resumo:
This paper describes multiple field-coupled simulations and device characterization of fully CMOS-MEMS-compatible smart gas sensors. The sensor structure is designated for gas/vapour detection at high temperatures (>300 °C) with low power consumption, high sensitivity and competent mechanic robustness employing the silicon-on-insulator (SOI) wafer technology, CMOS process and micromachining techniques. The smart gas sensor features micro-heaters using p-type MOSFETs or polysilicon resistors and differentially transducing circuits for in situ temperature measurement. Physical models and 3D electro-thermo-mechanical simulations of the SOI micro-hotplate induced by Joule, self-heating, mechanic stress and piezoresistive effects are provided. The electro-thermal effect initiates and thus affects electronic and mechanical characteristics of the sensor devices at high temperatures. Experiments on variation and characterization of micro-heater resistance, power consumption, thermal imaging, deformation interferometry and dynamic thermal response of the SOI micro-hotplate have been presented and discussed. The full integration of the smart gas sensor with automatically temperature-reading ICs demonstrates the lowest power consumption of 57 mW at 300 °C and fast thermal response of 10 ms. © 2008 IOP Publishing Ltd.
Discriminative language model adaptation for Mandarin broadcast speech transcription and translation
Resumo:
This paper investigates unsupervised test-time adaptation of language models (LM) using discriminative methods for a Mandarin broadcast speech transcription and translation task. A standard approach to adapt interpolated language models to is to optimize the component weights by minimizing the perplexity on supervision data. This is a widely made approximation for language modeling in automatic speech recognition (ASR) systems. For speech translation tasks, it is unclear whether a strong correlation still exists between perplexity and various forms of error cost functions in recognition and translation stages. The proposed minimum Bayes risk (MBR) based approach provides a flexible framework for unsupervised LM adaptation. It generalizes to a variety of forms of recognition and translation error metrics. LM adaptation is performed at the audio document level using either the character error rate (CER), or translation edit rate (TER) as the cost function. An efficient parameter estimation scheme using the extended Baum-Welch (EBW) algorithm is proposed. Experimental results on a state-of-the-art speech recognition and translation system are presented. The MBR adapted language models gave the best recognition and translation performance and reduced the TER score by up to 0.54% absolute. © 2007 IEEE.
Resumo:
In speech recognition systems language model (LMs) are often constructed by training and combining multiple n-gram models. They can be either used to represent different genres or tasks found in diverse text sources, or capture stochastic properties of different linguistic symbol sequences, for example, syllables and words. Unsupervised LM adaptation may also be used to further improve robustness to varying styles or tasks. When using these techniques, extensive software changes are often required. In this paper an alternative and more general approach based on weighted finite state transducers (WFSTs) is investigated for LM combination and adaptation. As it is entirely based on well-defined WFST operations, minimum change to decoding tools is needed. A wide range of LM combination configurations can be flexibly supported. An efficient on-the-fly WFST decoding algorithm is also proposed. Significant error rate gains of 7.3% relative were obtained on a state-of-the-art broadcast audio recognition task using a history dependently adapted multi-level LM modelling both syllable and word sequences. ©2010 IEEE.
Resumo:
In this communication, we describe a new method which has enabled the first patterning of human neurons (derived from the human teratocarcinoma cell line (hNT)) on parylene-C/silicon dioxide substrates. We reveal the details of the nanofabrication processes, cell differentiation and culturing protocols necessary to successfully pattern hNT neurons which are each key aspects of this new method. The benefits in patterning human neurons on silicon chip using an accessible cell line and robust patterning technology are of widespread value. Thus, using a combined technology such as this will facilitate the detailed study of the pathological human brain at both the single cell and network level. © 2010 Elsevier B.V.