首页>
外国专利>
PROBABILISTIC TRAINING FOR BINARY NEURAL NETWORKS
PROBABILISTIC TRAINING FOR BINARY NEURAL NETWORKS
展开▼
机译:二进制神经网络的概率训练
展开▼
页面导航
摘要
著录项
相似文献
摘要
A method (100) for training an artificial neural network (1) that is at least partially implemented as a binary neural network (2), comprising• receiving (110) training input values (11) for the inputs xi of the artificial neural network (1),• processing (120) training input values (11) to obtain output values (12),• applying (130), to obtained output values (12), a loss function (13);• updating (140) training weights wlj and branching back (150) to processing (120) until rating (13a) by the loss function (13) meets a predetermined termination criterion (160); and• configuring (170) actual artificial neural network (1, 2) according to finally obtained training weights wlj,wherein processing (120) comprises:• determining (121), for each neuron I in the layer (21-23), a distribution Al of pre-activations al using a distribution Wlj of weights wlj; and• determining (122) a distribution Hl of output values hl by applying, to distribution Al, predetermined thresholding function g.
展开▼
机译:一种用于训练至少部分地被实现为二进制神经网络(2)的人工神经网络(1)的方法(100),包括•接收(110)人工神经网络(1)的输入x i Sub>的训练输入值(11),•处理(120)训练输入值(11),以获得输出值(12),•将(130)损失函数(13)应用于获得的输出值(12);·更新(140)训练权重w lj Sub>,并分支(150)到处理(120),直到损失函数(13)的评分(13a)满足预定的终止标准(160);和•根据最终获得的训练权重w lj Sub>配置(170)实际的人工神经网络(1、2),其中处理(120)包括:•使用分布W 为层(21-23)中的每个神经元I确定(121)预激活a l Sub>的分布A l Sub>权重为w lj Sub>的lj Sub>;和·通过将预定阈值函数g应用于分布A l Sub>来确定(122)输出值h l Sub>的分布H l Sub>。
展开▼