cifar-100数据集

线性神经元是深度神经网络的最基本组成部分。 可以如下图所示。 在这里，X = {x1, ... ,xn}代表输入向量， w[i]是神经元的权重。 给定一个包含一组输入目标值对的训练集，线性神经元尝试学习一种线性变换，该变换可以将输入向量映射到相应的目标值。 基本上，线性神经元通过线性函数W^T x = y近似输入输出关系：

Linear neurons are the most fundamental building blocks of deep neural networks, as illustrated below. Here, X = {x₁, ..., xₙ} represents the input vector, and w[i] denotes the weights of the neuron. Given a training set consisting of input-target pairs, linear neurons attempt to learn a linear transformation that maps input vectors to their corresponding target values. Essentially, linear neurons approximate the input-output relationship via the linear function W^T x = y: