Two discriminative methods for solving tone problems in Mandarin speech recognition are presented. First, discriminative training on the HMM （hidden Markov model） based tone models is proposed. Then an integration t...Two discriminative methods for solving tone problems in Mandarin speech recognition are presented. First, discriminative training on the HMM （hidden Markov model） based tone models is proposed. Then an integration technique of tone models into a large vocabulary continuous speech recognition system is presented. Discriminative model weight training based on minimum phone error criteria is adopted aiming at optimal integration of the tone models. The extended Baum Welch algorithm is applied to find the model-dependent weights to scale the acoustic scores and tone scores. Experimental results show that tone recognition rates and continuous speech recognition accuracy can be improved by the discriminatively trained tone model. Performance of a large vocabulary continuous Mandarin speech recognition system can be further enhanced by the discriminatively trained weight combinations due to a better interpolation of the given models.展开更多
In the field of the water resources, hydrologic models have been used to assess water quality performance of complex watersheds and river basins. Hydrologic models can provide essential information for making decision...In the field of the water resources, hydrologic models have been used to assess water quality performance of complex watersheds and river basins. Hydrologic models can provide essential information for making decisions on sustainable management system of water resources within watersheds. The main objective of this study was to validate the performance of the Soil and Water Assessment Tool （SWAT） and the feasibility of using this model as a simulator of runoff at a catchment scale in semi-arid area in Northwestern Tunisia. Calibration and validation of the model output were performed by comparing predicted runoff with corresponding measurements from the Sarrath outlet for the periods 1990-1995 for calibration and 2000-2005 for validation. The time series for the years 1996-1999 showed discrepancies between the measured rainfall and the observed runoff indicating errors due to either the observations or to a dysfunction in the equipments. Sensitivity analysis shows that sensitive parameters for the simulation of discharge include curve number, soil evaporation compensation factor, depth of water in shallow aquifer and slope of subbasin. Statistical comparisons between monthly simulated results and observed data for the calibration period gave a reasonable agreement with a coefficient of determination （R2） greater than 0.75 and Nash-Sutcliffe Coefficient （NSE） equal to 0.72. These values were respectively 0.70 and 0.64 for validation period. Overall, the SWAT model has the capability to predict runoff within a complex semi-arid catchment.展开更多
In this article, the Bayes linear unbiased estimation （BALUE） of parameters is derived for the partitioned linear model. The superiorities of the BALUE over ordinary least square estimator （LSE） are studied in ter...In this article, the Bayes linear unbiased estimation （BALUE） of parameters is derived for the partitioned linear model. The superiorities of the BALUE over ordinary least square estimator （LSE） are studied in terms of the Bayes mean square error matrix （BMSEM） criterion and Pitman closeness （PC） criterion.展开更多
Based on remote sensing data of the Yangtze River Delta(YRD) in the years of 1991, 2001 and 2008, the paper built an index system of land use potential restraint factors in YRD, according to geological condition, terr...Based on remote sensing data of the Yangtze River Delta(YRD) in the years of 1991, 2001 and 2008, the paper built an index system of land use potential restraint factors in YRD, according to geological condition, terrain condition, water area, natural reserve area and basic farmland, and evaluated construction land potential based on the platform of GIS spatial analysis model. The results showed that:(1) the construction land increased rapidly since 1991 and reached 24,951.21 km2 in 2008, or 21.27% of the total area. Among all the cities in the YRD, Shanghai took the greatest percentage, followed by Jiangsu and Zhejiang. Spatially, areas where government departments are located became the growth center of construction land. Prefecture-level cities were the fastest growth region and the changing trend showed circle layered characteristics and significant increase with Shanghai and Suzhou as the core.(2) The higher the quality of construction land potentials(CLP), the smaller the number of CLP units. High sensitive area accounted for the largest percentage(40.14%) among all types of constraint regions and this was followed by medium sensitive region(31.53%) of the whole region.(3) The comprehensive CLP in the YRD was 24,989.65 km2, or 21.76% of the total YRD. The land use potential showed spatial distribution imbalance. CLP of Zhejiang was obviously larger than that of Jiangsu. CLP was insufficient in regional central city. Moreover, CLP in the YRD formed a circle layered spatial pattern that increasingly expanded centered in prefecture-level cities. Low potential area expanded from north to south. High potential area was mainly located in south YRD. Areas with zero potential in the YRD formed a northwest-southeast "Y-shaped" spatial pattern in north Hangzhou Bay.(4) CLP per capita in YRD was 0.045 ha/person and also unevenly distributed. Some 25.57% of the study units at county level nearly had no construction land and 8.24% of the units had CLP per capita below the national average level. CLP per capita in less than 25% of the county-level units was larger than the YRD average level, which were mainly located in Zhejiang. Therefore, research on the construction potential area in YRD was favorable for analysis of the development status and potential space of this region under the background of rapid urbanization and industrialization.展开更多
文摘Two discriminative methods for solving tone problems in Mandarin speech recognition are presented. First, discriminative training on the HMM （hidden Markov model） based tone models is proposed. Then an integration technique of tone models into a large vocabulary continuous speech recognition system is presented. Discriminative model weight training based on minimum phone error criteria is adopted aiming at optimal integration of the tone models. The extended Baum Welch algorithm is applied to find the model-dependent weights to scale the acoustic scores and tone scores. Experimental results show that tone recognition rates and continuous speech recognition accuracy can be improved by the discriminatively trained tone model. Performance of a large vocabulary continuous Mandarin speech recognition system can be further enhanced by the discriminatively trained weight combinations due to a better interpolation of the given models.
文摘In the field of the water resources, hydrologic models have been used to assess water quality performance of complex watersheds and river basins. Hydrologic models can provide essential information for making decisions on sustainable management system of water resources within watersheds. The main objective of this study was to validate the performance of the Soil and Water Assessment Tool （SWAT） and the feasibility of using this model as a simulator of runoff at a catchment scale in semi-arid area in Northwestern Tunisia. Calibration and validation of the model output were performed by comparing predicted runoff with corresponding measurements from the Sarrath outlet for the periods 1990-1995 for calibration and 2000-2005 for validation. The time series for the years 1996-1999 showed discrepancies between the measured rainfall and the observed runoff indicating errors due to either the observations or to a dysfunction in the equipments. Sensitivity analysis shows that sensitive parameters for the simulation of discharge include curve number, soil evaporation compensation factor, depth of water in shallow aquifer and slope of subbasin. Statistical comparisons between monthly simulated results and observed data for the calibration period gave a reasonable agreement with a coefficient of determination （R2） greater than 0.75 and Nash-Sutcliffe Coefficient （NSE） equal to 0.72. These values were respectively 0.70 and 0.64 for validation period. Overall, the SWAT model has the capability to predict runoff within a complex semi-arid catchment.
基金This research is supported by National Natural Science Foundation of China under Grant Nos. 10801123, 10801124 and 10771204, and the Knowledge Innovation Program of the Chinese Academy of Sciences under Grant No. KJCX3-SYW-S02.
文摘In this article, the Bayes linear unbiased estimation （BALUE） of parameters is derived for the partitioned linear model. The superiorities of the BALUE over ordinary least square estimator （LSE） are studied in terms of the Bayes mean square error matrix （BMSEM） criterion and Pitman closeness （PC） criterion.
基金National Natural Youth Science Foundation of China,No.41201168
文摘Based on remote sensing data of the Yangtze River Delta(YRD) in the years of 1991, 2001 and 2008, the paper built an index system of land use potential restraint factors in YRD, according to geological condition, terrain condition, water area, natural reserve area and basic farmland, and evaluated construction land potential based on the platform of GIS spatial analysis model. The results showed that:(1) the construction land increased rapidly since 1991 and reached 24,951.21 km2 in 2008, or 21.27% of the total area. Among all the cities in the YRD, Shanghai took the greatest percentage, followed by Jiangsu and Zhejiang. Spatially, areas where government departments are located became the growth center of construction land. Prefecture-level cities were the fastest growth region and the changing trend showed circle layered characteristics and significant increase with Shanghai and Suzhou as the core.(2) The higher the quality of construction land potentials(CLP), the smaller the number of CLP units. High sensitive area accounted for the largest percentage(40.14%) among all types of constraint regions and this was followed by medium sensitive region(31.53%) of the whole region.(3) The comprehensive CLP in the YRD was 24,989.65 km2, or 21.76% of the total YRD. The land use potential showed spatial distribution imbalance. CLP of Zhejiang was obviously larger than that of Jiangsu. CLP was insufficient in regional central city. Moreover, CLP in the YRD formed a circle layered spatial pattern that increasingly expanded centered in prefecture-level cities. Low potential area expanded from north to south. High potential area was mainly located in south YRD. Areas with zero potential in the YRD formed a northwest-southeast "Y-shaped" spatial pattern in north Hangzhou Bay.(4) CLP per capita in YRD was 0.045 ha/person and also unevenly distributed. Some 25.57% of the study units at county level nearly had no construction land and 8.24% of the units had CLP per capita below the national average level. CLP per capita in less than 25% of the county-level units was larger than the YRD average level, which were mainly located in Zhejiang. Therefore, research on the construction potential area in YRD was favorable for analysis of the development status and potential space of this region under the background of rapid urbanization and industrialization.
