查询词典 generalization

The penalty optimal brain surgeon is a post-training algorithm and it has extreme high complexity. The OBS oriented compute model implemented neural network training and OBS pruning simultaneously, by taking the OBS pruning case as a penalty term of neural network objective functions based on optimized structure of the regularization method included in the neural network training process. It both maintained the OBS's accuracy and had regularization method's high efficiency. Raised the generalization of neural network model.

最优脑外科过程是一种训练后网络剪枝算法，计算的复杂度非常高，通过把剪枝条件以惩罚项的形式纳入神经网络的训练目标函数中，把正则化方法的结构优化蕴涵于网络训练过程，构建面向最优脑外科过程的计算模型，实现网络训练过程和最优脑外科过程并行剪枝，既保持了最优脑外科过程的准确性，又具有正则化的高效性，提高了神经网络模型的泛化性能。

The theory of generalized metric properties and covering properties stems from a generalization of metrizability and compactness, in which a great many problems involve a research for certain point-countable covers.

作为对可度量性与紧性一般化而形成的广义度量理论与覆盖性质理论中的许多问题涉及到对确定的点可数。。。

In [33] and [34], a nice generalization of minimal Lagrangian submanifolds called Hamiltonian-minimal submanifold was introduced and investigated by Y.G.

Oh在[33]和[34]中最先引进的哈密尔顿极小拉格朗日子流形，此类子流形式是对极小拉格朗日子流形一种很好的推广，同时他对此类子流形作了相应的研究。

In chapter 4 , with S-KKM mapping , a new Ky Fan\'s minimax inequality is proved , the locally convex H-space with uniform structure is introduced , s-HKKM mapping is defined as the generalization of KKM mapping, with the H-convexity of Hausdorff locally convex H-space , a new fixed point theorem for the compact andclosed mapping belonging to the class s-HKKM is established ;with local intersection property of muti-function ,K-H-quasi-convexity of mapping , connectedness of set and condensing correspondences, some new vector quasi-variational inequalities are established , respectively .

第四章利用S-KKM映射，建立了新的Ky Fan极小极大不等式；定义了带有一致结构的局部凸H-空间，利用Hausdorff局部凸H-空间的H-凸性，对于属于s-HKKM中紧的闭映射建立了新的不动点定理，利用对应的局部交性质、映射的K-拟-凸性、集合的连通性以及φ-condensing对应分别建立了相应的拟向量变分不等式。

Firstly, the generalization of Fan-Ha section theorem and a general vector variational inequality without convexity assumption and minimax theorem of vector-valued function are obtained. Then, the quasi-montone vector variational inequality problem is concerned. Cx-quasi-monotone operator is defined in topological vector space, inner point of a closed convex set K is introduced, the relation between inner point and relative algebraic interior point is given, an existence result for quasi-monotone vector variational inequality is obtained.

第三章主要研究了向量变分不等式和极小极大定理(来源：3282AB83C论文网www.abclunwen.com)，建立了广义的Fan-Ha截口定理、新的向量变分不等式与极小极大定理，并在拓扑向量空间中定义了C_x-拟单调算子，引入了闭凸集K的inner点，给出了inner点与相对代数内点的关系，利用innK_c代替K的拓扑内部，建立了新的拟单调向量变分不等式。

Moreover, a method minimizing both model fitting capabilities and generalization as the objective function is offered to choose the proper parameters of support vector regression and facilitate the application.

给出了一种以最小化支持向量回归的拟合能力和泛化能力为目标函数来选择支持向量回归机参数的方法，以方便该方法的工程应用。

Based on SLT and Structure Risk Minimum principle, Support Vector Machine is not only simple in structure, but also has good generalization ability.

基于统计学习理论和结构风险最小化原则的支持向量机不仅模型结构简单，而且具有良好的泛化能力。

Since weak Hopf algebra is a generalization of monoid algebra, it is easy to ask what about the weak Hopf algebra?

而且，类似于[L8]中的相关结论，我们将给出交换情况下的半格分次弱Hopf代数的G-量子偶的分解和半单性的讨论。

Although the quantum double of a finite Clifford monoid is indeed a generalization of the quantum double of a finite group, the quantum doubles in [L2] can not usually be regarded as

作为特殊情况，非交换的半格分次弱Hopf代数的量子偶同样可以构造出来。作为群和非交换非余交换Hopf代数的量子偶的构

NET effectively balance "function of generalization" and "implementation efficiency," ADO.

NET有效地平衡"功能的泛化"和"执行效率"，以及如何用ADO。

Molecular Distillation is a kind of high vacuum distillation method, which is suitable for the separation of high boiling, heat sensitivity and viscidity products.

分子蒸馏技术是一种在高真空条件下进行的连续蒸馏过程，适合于分离高沸点、热敏性及具有生物活性的混合物。

They are to the Noughties what yuppies were to the Eighties .

那在六十年代是时髦的，但在八十年代肯定是过时了