Speaker separation in complex acoustic environment is one of challenging tasks in speech separation.In practice,speakers are very often unmoving or moving slowly in normal communication.In this case,the spatial featur...Speaker separation in complex acoustic environment is one of challenging tasks in speech separation.In practice,speakers are very often unmoving or moving slowly in normal communication.In this case,the spatial features among the consecutive speech frames become highly correlated such that it is helpful for speaker separation by providing additional spatial information.To fully exploit this information,we design a separation system on Recurrent Neural Network(RNN)with long short-term memory(LSTM)which effectively learns the temporal dynamics of spatial features.In detail,a LSTM-based speaker separation algorithm is proposed to extract the spatial features in each time-frequency(TF)unit and form the corresponding feature vector.Then,we treat speaker separation as a supervised learning problem,where a modified ideal ratio mask(IRM)is defined as the training function during LSTM learning.Simulations show that the proposed system achieves attractive separation performance in noisy and reverberant environments.Specifically,during the untrained acoustic test with limited priors,e.g.,unmatched signal to noise ratio(SNR)and reverberation,the proposed LSTM based algorithm can still outperforms the existing DNN based method in the measures of PESQ and STOI.It indicates our method is more robust in untrained conditions.展开更多
AIM To evaluate the accuracy of the elastography score combined to the strain ratio in the diagnosis of solid pancreatic lesions(SPL). METHODS A total of 172 patients with SPL identified by endoscopic ultrasound were ...AIM To evaluate the accuracy of the elastography score combined to the strain ratio in the diagnosis of solid pancreatic lesions(SPL). METHODS A total of 172 patients with SPL identified by endoscopic ultrasound were enrolled in the study to evaluate the efficacy of elastography and strain ratio in differentiating malignant from benign lesions. The semi quantitative score of elastography was represented by the strain ratio method. Two areas were selected, area(A) representing the region of interest and area(B) representing the normal area. Area(B) was then divided by area(A). Sensitivity, specificity, positive predictive value(PPV), negative predictive value(NPV), and accuracy were calculated by comparing diagnoses made by elastography, strain ratio and final diagnoses.RESULTS SPL were shown to be benign in 49 patients and malignant in 123 patients. Elastography alone had a sensitivity of 99%, a specificity of 63%, and an accuracy of 88%, a PPV of 87% and an NPV of 96%. The best cut-off level of strain ratio to obtain the maximal area under the curve was 7.8 with a sensitivity of 92%, specificity of 77%, PPV of 91%, NPV of 80% and an accuracy of 88%. Another estimated cut off strain ratio level of 3.8 had a higher sensitivity of 99% and NPV of 96%, but with less specificity, PPV and accuracy 53%, 84% and 86%, respectively. Adding both elastography to strain ratio resulted in a sensitivity of 98%, specificity of 77%, PPV of 91%, NPV of 95% and accuracy of 92% for the diagnosis of SPL. CONCLUSION Combining elastography to strain ratio increases the accuracy of the differentiation of benign from malignant SPL.展开更多
The arrival times of first teleseismic phases are difficult to be measured precisely because of slowly and gradually changed onsets and weak amplitudes. The arrival times measured manually are usually behind the real ...The arrival times of first teleseismic phases are difficult to be measured precisely because of slowly and gradually changed onsets and weak amplitudes. The arrival times measured manually are usually behind the real ones. In this paper, using the ratio method of fixed scale wavelet transformations improved by us, the arrival times for the first arrival phases (such as P and PKIKP) at the teleseismic and far-teleseismic distances were measured. The results are reasonable and reliable based on the analysis and discussion of the reliabilities and errors.展开更多
The present study was to investigate the effects of feeding time on growth of the Chinese soft-shelled turtle. The experiment was carried out with a photoperiod of 12L （light）： 12D （dark） with lights on at 06：00...The present study was to investigate the effects of feeding time on growth of the Chinese soft-shelled turtle. The experiment was carried out with a photoperiod of 12L （light）： 12D （dark） with lights on at 06：00 and off at 18：00. Turtles were maintained at a temperature of 32 - 0.2 ℃ in tanks throughout the length of the experiment. The turtles in group 1 to group 6 were fed respectively at 00：00, 04：00, 08：00, 12：00, 16：00 and 20：00 with 60 turtles each （Initial body weight 88.27 - 0.09 g）. Acrophases of postprandial RNA/DNA ratio in liver in each group was shown between 5h and 7h after feeding. A positive linear correlation could be seen between specific growth rate （SGR） and RNA/ DNA：SGR=t.1586RNA/DNA-0.7097 （r = 0.9328, P = 0.0066）. The results indicated that the values at the acrophases of about 6h after feeding might be used as an instantaneous growth index in the Chinese soft-shelled turtle. The turtles in group 1 grew better than group 3, group 4, group 5, group 6, because they ate more, but they ate more and grew slower than group 2, whose feed conversion rate was also higher. Meanwhile, the SGR and feeding rate of turtles fed at 12：00 were the lowest from the six groups （P 〈 0.05）. Turtles fed in group 1, group 2 and group 6 developed more heavy final body weight （FBW）, higher feeding rate and SGR than the other three groups. This probably suggested that turtles fed in scotophase grew better than that fed at photophase in total.展开更多
We demonstrate here that global-scale determination of a key ionospheric parameter,the peak height of the F_(2)region(h_(m)F_(2)),can be obtained by making a simple ratio measurement of the atomic oxygen 130.4 and 135...We demonstrate here that global-scale determination of a key ionospheric parameter,the peak height of the F_(2)region(h_(m)F_(2)),can be obtained by making a simple ratio measurement of the atomic oxygen 130.4 and 135.6 nm emissions in the far-ultraviolet nightglow with a nadir-viewing system such as a pair of photometers suitable for flight on a CubeSat.We further demonstrate that measurements from an altitude that is within the typical range of nighttime h_(m)F_(2)250−450 km can provide the ratios that are needed for retrieval of the h_(m)F_(2).Our study is conducted mostly through numerical simulations by using radiative transfer models of the two emissions coupled with empirical models of the atmosphere and ionosphere.Modeling results show that the relationship between the h_(m)F_(2)and the intensity ratio is sensitive to the altitude from which the emissions are observed,primarily because of the distinctly different degrees of resonant scattering of the two emissions in the atmosphere.A roughly quadratic relationship can be established for observations from an orbit of~400 km,which enables h_(m)F_(2)retrieval.Parametric analysis indicates that the relationship can be affected by the ambient atmospheric conditions through resonant scattering and O2 absorption.For typical nighttime conditions with h_(m)F_(2)250−450 km,retrieval of the h_(m)F_(2)from synthetic observations shows that the typical errors are only a few kilometers(up to~20 km),depending on the accuracy of the ambient conditions predicted by the empirical models.Our findings pave the way for use of the 130.4/135.6 nm intensity ratios for global-scale monitoring of the nighttime ionosphere at mid to low latitudes.展开更多
In this paper, we propose a discrete ratio-dependent predator-prey system. The stability of the fixed points of this model is studied. At the same time, it is shown that the discrete model undergoes fold bifurcation a...In this paper, we propose a discrete ratio-dependent predator-prey system. The stability of the fixed points of this model is studied. At the same time, it is shown that the discrete model undergoes fold bifurcation and flip bifurcation by using bifurcation theory and the method of approximation by a flow. Numerical simulations are presented not only to demonstrate the consistence with our theoretical analyses, but also to exhibit the complex dynamical behaviors, such as the cascade of period-doubling bifurcation in period-2 and the chaotic sets. The Maximum Lyapunov exponents are numerically computed to confirm further the complexity of the dynamical behaviors. These results show that the direct discrete method has more rich dynamic behaviors than the discrete model obtained by Euler method.展开更多
基金This work is supported by the National Nature Science Foundation of China(NSFC)under Grant Nos.61571106,61501169,41706103the Fundamental Research Funds for the Central Universities under Grant No.2242013K30010.
文摘Speaker separation in complex acoustic environment is one of challenging tasks in speech separation.In practice,speakers are very often unmoving or moving slowly in normal communication.In this case,the spatial features among the consecutive speech frames become highly correlated such that it is helpful for speaker separation by providing additional spatial information.To fully exploit this information,we design a separation system on Recurrent Neural Network(RNN)with long short-term memory(LSTM)which effectively learns the temporal dynamics of spatial features.In detail,a LSTM-based speaker separation algorithm is proposed to extract the spatial features in each time-frequency(TF)unit and form the corresponding feature vector.Then,we treat speaker separation as a supervised learning problem,where a modified ideal ratio mask(IRM)is defined as the training function during LSTM learning.Simulations show that the proposed system achieves attractive separation performance in noisy and reverberant environments.Specifically,during the untrained acoustic test with limited priors,e.g.,unmatched signal to noise ratio(SNR)and reverberation,the proposed LSTM based algorithm can still outperforms the existing DNN based method in the measures of PESQ and STOI.It indicates our method is more robust in untrained conditions.
文摘AIM To evaluate the accuracy of the elastography score combined to the strain ratio in the diagnosis of solid pancreatic lesions(SPL). METHODS A total of 172 patients with SPL identified by endoscopic ultrasound were enrolled in the study to evaluate the efficacy of elastography and strain ratio in differentiating malignant from benign lesions. The semi quantitative score of elastography was represented by the strain ratio method. Two areas were selected, area(A) representing the region of interest and area(B) representing the normal area. Area(B) was then divided by area(A). Sensitivity, specificity, positive predictive value(PPV), negative predictive value(NPV), and accuracy were calculated by comparing diagnoses made by elastography, strain ratio and final diagnoses.RESULTS SPL were shown to be benign in 49 patients and malignant in 123 patients. Elastography alone had a sensitivity of 99%, a specificity of 63%, and an accuracy of 88%, a PPV of 87% and an NPV of 96%. The best cut-off level of strain ratio to obtain the maximal area under the curve was 7.8 with a sensitivity of 92%, specificity of 77%, PPV of 91%, NPV of 80% and an accuracy of 88%. Another estimated cut off strain ratio level of 3.8 had a higher sensitivity of 99% and NPV of 96%, but with less specificity, PPV and accuracy 53%, 84% and 86%, respectively. Adding both elastography to strain ratio resulted in a sensitivity of 98%, specificity of 77%, PPV of 91%, NPV of 95% and accuracy of 92% for the diagnosis of SPL. CONCLUSION Combining elastography to strain ratio increases the accuracy of the differentiation of benign from malignant SPL.
基金National Natural Science Foundation of China (40074007).
文摘The arrival times of first teleseismic phases are difficult to be measured precisely because of slowly and gradually changed onsets and weak amplitudes. The arrival times measured manually are usually behind the real ones. In this paper, using the ratio method of fixed scale wavelet transformations improved by us, the arrival times for the first arrival phases (such as P and PKIKP) at the teleseismic and far-teleseismic distances were measured. The results are reasonable and reliable based on the analysis and discussion of the reliabilities and errors.
基金supported by the National Natural Science Foundation of China (No.31172085)
文摘The present study was to investigate the effects of feeding time on growth of the Chinese soft-shelled turtle. The experiment was carried out with a photoperiod of 12L （light）： 12D （dark） with lights on at 06：00 and off at 18：00. Turtles were maintained at a temperature of 32 - 0.2 ℃ in tanks throughout the length of the experiment. The turtles in group 1 to group 6 were fed respectively at 00：00, 04：00, 08：00, 12：00, 16：00 and 20：00 with 60 turtles each （Initial body weight 88.27 - 0.09 g）. Acrophases of postprandial RNA/DNA ratio in liver in each group was shown between 5h and 7h after feeding. A positive linear correlation could be seen between specific growth rate （SGR） and RNA/ DNA：SGR=t.1586RNA/DNA-0.7097 （r = 0.9328, P = 0.0066）. The results indicated that the values at the acrophases of about 6h after feeding might be used as an instantaneous growth index in the Chinese soft-shelled turtle. The turtles in group 1 grew better than group 3, group 4, group 5, group 6, because they ate more, but they ate more and grew slower than group 2, whose feed conversion rate was also higher. Meanwhile, the SGR and feeding rate of turtles fed at 12：00 were the lowest from the six groups （P 〈 0.05）. Turtles fed in group 1, group 2 and group 6 developed more heavy final body weight （FBW）, higher feeding rate and SGR than the other three groups. This probably suggested that turtles fed in scotophase grew better than that fed at photophase in total.
基金the National Natural Science Foundation of China through Grant 8206100245the Chinese Meteorological Administration through Grant FY-APP-ZX-2022.0222.
文摘We demonstrate here that global-scale determination of a key ionospheric parameter,the peak height of the F_(2)region(h_(m)F_(2)),can be obtained by making a simple ratio measurement of the atomic oxygen 130.4 and 135.6 nm emissions in the far-ultraviolet nightglow with a nadir-viewing system such as a pair of photometers suitable for flight on a CubeSat.We further demonstrate that measurements from an altitude that is within the typical range of nighttime h_(m)F_(2)250−450 km can provide the ratios that are needed for retrieval of the h_(m)F_(2).Our study is conducted mostly through numerical simulations by using radiative transfer models of the two emissions coupled with empirical models of the atmosphere and ionosphere.Modeling results show that the relationship between the h_(m)F_(2)and the intensity ratio is sensitive to the altitude from which the emissions are observed,primarily because of the distinctly different degrees of resonant scattering of the two emissions in the atmosphere.A roughly quadratic relationship can be established for observations from an orbit of~400 km,which enables h_(m)F_(2)retrieval.Parametric analysis indicates that the relationship can be affected by the ambient atmospheric conditions through resonant scattering and O2 absorption.For typical nighttime conditions with h_(m)F_(2)250−450 km,retrieval of the h_(m)F_(2)from synthetic observations shows that the typical errors are only a few kilometers(up to~20 km),depending on the accuracy of the ambient conditions predicted by the empirical models.Our findings pave the way for use of the 130.4/135.6 nm intensity ratios for global-scale monitoring of the nighttime ionosphere at mid to low latitudes.
文摘In this paper, we propose a discrete ratio-dependent predator-prey system. The stability of the fixed points of this model is studied. At the same time, it is shown that the discrete model undergoes fold bifurcation and flip bifurcation by using bifurcation theory and the method of approximation by a flow. Numerical simulations are presented not only to demonstrate the consistence with our theoretical analyses, but also to exhibit the complex dynamical behaviors, such as the cascade of period-doubling bifurcation in period-2 and the chaotic sets. The Maximum Lyapunov exponents are numerically computed to confirm further the complexity of the dynamical behaviors. These results show that the direct discrete method has more rich dynamic behaviors than the discrete model obtained by Euler method.
