1)  error averaging

1.
The algorithm adopts variable-scale searching and error averaging techniques to train the weight of the network,and(converts) weight training problem into optimization searching in multi parameters space.

2)  mean square analysis

3)  Error mean distribution

4)  rms percent error

5)  DMMSE(Difference minimum mean square error)

1.
Algorithms based on DMMSE(Difference minimum mean square error) criterion of adaptive multi users detector for short spreading sequences DS CDMA systems is proposed.

6)  mean square error

1.
Firstly,screen the descriptors using support vector machine regression(SVR) by leave-one-out method based on the minimum mean square error(MSE),get the optimal kernel and the corresponding retained descriptors.

2.
For the generalized linear model with aggregated data:Y=Xβ+u,Eu=0,Var(u)=σ2∑,this paper is built two kinds of biased estimators: ridge estimator β(k)and improved ridge estimator β(k)which are discussed some superiority to the estimators in the sense of mean square error.

3.
As to seemingly unrelated regression system,a new biased contracting estimator of the parameters is put forward,which is the combination of generalized ridge covariance-improved estimator and Stein estimator,and the good features of this estimator in mean square error is discussed.

 能量均分定理equipartition，theorem   在平衡态，物质分子能量分配的统计规律。在温度为T的平衡态，物质（气体、液体或固体）分子的每一个运动自由度都具有相同的平均动能，其中k是玻耳兹曼常量。用它可确定物质的内能和热容量。   如果气体分子的平动、转动、振动自由度分别为t、r、s，则相应的平均动能分别为，因分子内原子的振动除动能外还有势能，且两者在一个周期内的平均值相等，故对于每一个振动自由度，还有的平均势能。于是，分子的平均总能量为，对于理想气体，因分子间无相互作用，其内能就是上述分子平均能量与总分子数的乘积，只与温度有关，其热容量则与温度无关。   气体分子能量按自由度的均分通过分子间的频繁碰撞得以建立和维持。外界供给气体的能量，通过器壁与气体分子以及气体分子之间的碰撞分配到各自由度上。对于液体和固体，能量按自由度均分是通过分子间很强的相互作用而实现的。   能量均分定理在经典物理范围内适用，要求能量可以连续地变化，它的困难也正源于此。