代替

The method replaces the grads in gradient method with difference quotient and can be used to many cases.

然后根据此界的特点，提出了基于差商的参数优化算法，用差商代替导数的计算，具有较大的适用范围。

Through the comparison with other learning methods, the conclusion is drawn that the Powell method in which the derivative is replaced by the difference quotient is an eff...

通过与其他学习算法的比较，得出结论：其中用差商近似代替导数的Powel法是一种高效、快速的学习方法，其学习速率比带动量项的学习率自适应的BP算法高一个量级，而比DavidenFletcherPowel、BroydenFletcherGoldfarlShanno等变尺度方法以及其他非线性最小二乘法的稳定性要好得多。

Thirdly, in order to improve the learning efficiency and stability of feedforward neural network, a fast learning algorithm for neural networks based on CGM-OC approach is presented. Compared with other learning methods such as BP method, Broyden Flecher Goldfarl Shanno method and Powell method in which the derivative is replaced by the difference quotient etc, simulation results show that the proposed method is an efficient and fast one.

第三，为了提高前馈神经网络权值的学习效率及稳定性，提出一种基于正交校正共轭梯度优化方法的神经网络学习算法，通过与其它学习算法（如：BP算法、变尺度法、用差商近似代替导数的Powell法等）的比较，经仿真试验表明，本算法是一种高效、快速的学习算法。

The weight values in NLSM were adjusted automatically according to target data, which allowed the measured data with different dimensions or ranges to be treated at the same level, and thus the accuracy and stability of algorithm were improved accordingly. Derivative was replaced by the difference quotient in NLSM which further improved the algorithm speed.

非线性最小二乘法的权值根据目标数据自动调整，使得不同量纲和不同数值范围的观测数据可以在同一水平上进行，提高了计算的精度和稳定性；另一方面用差商近似代替导数加快了算法的收敛速度。

The derivation course and the discrete scheme of the first model-Perona-Malik model have been expatiated on emphasis. One item of the P-M model is omitted in the P-M's discrete scheme, and the gradient magnitude is replaced by the absolute value of difference quotient.

Perona-Malik给出的离散格式是一种显式离散格式，这一格式的连续形式省略了原模型中的一项，且用差商的绝对值代替了梯度模。

We use B-spline smoothing technique to smooth the characteristic function without changing the integral quantity and get a differentiable weight function.

我们用B样条磨光技术在不改变积分值的前提下磨光特征函数，用可微的权重函数代替特征函数，提高了采样的质量。

Beijing Institute of Technology, Beijing 100081,ChinaIn order to get the high output power, a RF excited diffusively cooled all-metal slab waveguide CO2 laser was reported, in which gas discharge region was constructed by two aluminum side walls and aluminum electrodes. The region dimension was 2 mm height×20 mm width×386 mm length, and the gas composition was CO2:N2:He:Xe=1:1:3:0.26. This technique replaced the old ceramic and metal sandwich waveguide technique.The optimum out power 127 W was achieved from the RF excited diffusively cooled all-metal slab waveguide CO2 laser with the longitudinal voltage uniformity distribution technique and scaling area technique. The electro-optic conversion efficiency was about 14%.

为了获得较高的激光输出功率，介绍了射频激励扩散冷却全金属板条波导CO2激光器，放电区域由左右两个铝合金壁和上下两个铝合金电极构成，放电区域高2 mm，宽20 mm，长386 mm，工作气体混合比CO2：N2：He：Xe=1：1：3：0.26，该技术代替了过去的金属陶瓷结构，结合纵向电压均匀分布技术和面增比技术，获得了127 W的激光输出功率，光电转换效率高达14%。

Take my place at dim sum with my friends.

代替我去和我的朋友们喝早茶。

In order to reduce these artifacts which may occur in the phoneme pontes of the converted speech if the final part of a Chinese character or a syllable is a diphthong or a tri-phthong, we select a main vowel for each final to warp the spectrum in a longer segment: a final.

为了减轻基于音素的声音变换系统合成的声音频谱出现的很多不连续的现象，我们提出为每个韵母选取一个主要元音音素来代替对应的整个韵母部分，这样可以将语音频谱按音节分割成一些较长的频谱段，减少变换后频谱的不连续现象。

This paper chooses a normal mode helical antenna to replace the short dipole antenna to improve efficiency.

本文采用法向模螺旋天线代替以前采用的短偶极子天线以提高效率。

The split between the two groups can hardly be papered over.

这两个团体间的分歧难以掩饰。

This approach not only encourages a greater number of responses, but minimizes the likelihood of stale groupthink.

这种做法不仅鼓励了更多的反应，而且减少跟风的可能性。