期刊文献+
共找到4篇文章
< 1 >
每页显示 20 50 100
基于ASP-SERes2Net的说话人识别算法 被引量:1
1
作者 令晓明 陈鸿雁 +1 位作者 张小玉 张真 《北京工业大学学报》 CAS 北大核心 2025年第1期42-50,共9页
为提升说话人识别的特征提取能力,解决在噪声环境下识别率低的问题,提出一种基于残差网络的说话人识别算法——ASP-SERes2Net。首先,采用梅尔语谱图作为神经网络的输入;其次,改进Res2Net网络的残差块,并且在每个残差块后引入压缩激活(sq... 为提升说话人识别的特征提取能力,解决在噪声环境下识别率低的问题,提出一种基于残差网络的说话人识别算法——ASP-SERes2Net。首先,采用梅尔语谱图作为神经网络的输入;其次,改进Res2Net网络的残差块,并且在每个残差块后引入压缩激活(squeeze-and-excitation,SE)注意力模块;然后,用注意力统计池化(attention statistics pooling,ASP)代替原来的平均池化;最后,采用附加角裕度的Softmax(additive angular margin Softmax,AAM-Softmax)对说话人身份进行分类。通过实验,将ASP-SERes2Net算法与时延神经网络(time delay neural network,TDNN)、ResNet34和Res2Net进行对比,ASP-SERes2Net算法的最小检测代价函数(minimum detection cost function,MinDCF)值为0.0401,等误率(equal error rate,EER)为0.52%,明显优于其他3个模型。结果表明,ASP-SERes2Net算法性能更优,适合应用于噪声环境下的说话人识别。 展开更多
关键词 说话人识别 梅尔语谱图 Res2Net 压缩激活(squeeze-and-excitation SE)注意力模块 注意力统计池化(attention statistics pooling ASP) 附加角裕度的Softmax(additive angular margin Softmax AAM-Softmax)
在线阅读 下载PDF
基于多重网格控制体方法的皮米磁头气膜压强求解 被引量:4
2
作者 阮海林 王玉娟 +1 位作者 段传林 陈云飞 《传感技术学报》 CAS CSCD 北大核心 2006年第05B期2260-2263,共4页
磁头磁盘系统的发展越来越快,在磁头的开发设计阶段,就需要对磁头的飞行姿态作出快速、准确的数值预测.本文提出一种计算效率高的、基于叠加修正策略的多重网格控制体方法,解决轴承数很大和导轨形状复杂的气膜承载问题.运用这种方法求解... 磁头磁盘系统的发展越来越快,在磁头的开发设计阶段,就需要对磁头的飞行姿态作出快速、准确的数值预测.本文提出一种计算效率高的、基于叠加修正策略的多重网格控制体方法,解决轴承数很大和导轨形状复杂的气膜承载问题.运用这种方法求解Pico磁头(30%,1.25mm×1.0mm,飞高7nm)的气膜承载压强,相比较单网格方法,多重网格方法求解性能提高显著. 展开更多
关键词 磁头磁盘系统 ACM(additive CORRECTION multigrid)方法 皮米磁头 气膜承载压强
在线阅读 下载PDF
Detection and Analysis of High Temperature Sensitivity of TGMS Lines in Rice Using AMMI Model 被引量:4
3
作者 FULi-zhong XUEQing-zhong 《Agricultural Sciences in China》 CAS CSCD 2004年第9期671-677,共7页
With the AMMI (additive main effects and multiplicative interaction) analysis model, thedetermination of the sensitivity to temperature among different TGMS (thermo-sensitivegenic male sterile) lines was performed. To... With the AMMI (additive main effects and multiplicative interaction) analysis model, thedetermination of the sensitivity to temperature among different TGMS (thermo-sensitivegenic male sterile) lines was performed. To assess the genetic differences due to hightemperature stress at the fertility-sensitive stage (10-20d before heading), sevengenotypes (six TGMS lines and the control Pei-Ai64S) were grown from May 4 at sevendifferent stages with 10d intervals. The temperatures at the fertility-sensitive stagesinvolved twelve levels from<20 to>℃ under the regime natural conditions in Hangzhou,China. There was considerable variation in pollen fertility among genotypes in responseto high temperature. Five genotypes identified as TGMS lines as their percentages offertile pollens were lower than or close to that of the control except for the unstableline RTS19 (V6). When the temperatures at the fertility-sensitive stage were at Ⅰ-Ⅳ,Ⅴ-Ⅵ and Ⅶ-Ⅻ, the percentages of fertile pollens varied in the ranges of 46.46-48.49%,19.62-22.79% and 3.49-5.87%, respectively. The critical temperatures of sterility andfertility in the five TGMS lines were 25.1 and 23.0℃, respectively. Considering theamounts and directions of main effect and their IPCA (interaction principal componentsanalysis), we can classify the lines and temperature levels into different groups, anddescribe the characteristics of genotypetemperature interaction, offering the informationand tools for the development and utility of thermo-sensitive male sterile lines.Several TGMS rice lines with their reproductive sensitivity to high temperature that canbe screened using the AMMI model may add valuable germplasm to the breeding program ofhybrid rice. 展开更多
关键词 RICE AMMI (additive main effects and multiplicative interaction) TGMS (thermo- sensitive genic male sterile) FERTILITY
在线阅读 下载PDF
Genotype×year interaction of pod and seed mass and stability of Pongamia pinnata families in a semi-arid region
4
作者 G.R.Rao B.Sarkar +3 位作者 B.M.K.Raju P.Sathi Reddy A.V.M.Subba Rao Jessie Rebecca 《Journal of Forestry Research》 SCIE CAS CSCD 2020年第4期1333-1346,共14页
Sixteen pongamia families were evaluated in a field experiment for eight consecutive years in dryland conditions to identify stable,high-yielding families.The trial was conducted in a randomized complete block design ... Sixteen pongamia families were evaluated in a field experiment for eight consecutive years in dryland conditions to identify stable,high-yielding families.The trial was conducted in a randomized complete block design with three replications.Each family,consisting of nine trees per replication,was planted at a spacing of3 m x 3 m.Yield stability was analyzed using(1)Eberhart and Russel’s regression coefficient(β_i)and deviation from regression(S_d^2),(2)Wrike’s ecovalence(W_i);(3)Shukla stability variance(σ_i^2);and(4)Piepho and Lotito’s stability index(L_i).Families were also analyzed for adaptability and stability using AMMI and GGE biplots graphical methods.The study revealed significant variances due to family and family x year interaction for pod and seed yield.Families performed differently and ranked differently across years.The performance of families was influenced by both genetic factor and environmental conditions in different years.Among families tested,TNMP20,Acc14,TNMP14 and Acc30 were high yielders for pods,and Acc14,Acc30,TNMP6,RAK19 and TNMP14 were high for seed yield.According to the Eberhart and Russell model,Acc30,TNMP14 and TNMP3 were stable across years.In the graphical view of family x year interaction based on AMMI methods,TNMP3,TNMP4 and TNMP14 had greater stability with moderate seed yield,and Acc14 and Acc30 had moderate stability with high seed yield.On the other hand,GGE biplots revealed Acc14,Acc30 and TNMP14 as high yielders with moderate stability.AMMI and GGE biplots were able to capture nonlinear parts of the family x year interaction that were not be captured by the Eberhart and Russel model while also identifying stable families.Based on different methodologies,Acc14,Acc30 and TNMP14 were identified as high yielding and stable families for promoting pongamia cultivation as a biofuel crop for semi-arid regions. 展开更多
关键词 BIOFUEL Pongamia Genetic diversity STABILITY AMMI (additive main effects multiplicative interaction) GGE biplots Multi-year trial SVD(singular value decomposition)
在线阅读 下载PDF
上一页 1 下一页 到第
使用帮助 返回顶部