- matlab 对数据集的默认组织方式是,X∈Rd×N
- d:行数,表示特征向量的长度;
- N:列数,表示样本的数目;
1. 模型、预测、mse
% 加载 matlab 内置数据到内存
X = abalone_dataset;
% 模型定义
ae = trainAutoencoder(X);
% 训练集上的预测,对于自编码器而言,就是重构;
X_rec = predict(ae, X);
% 损失函数
mse_loss = mse(X-X_rec);
2. integer labels ⇒ categorical labels
>> integer_labels = randi([0, 9], 10, 1);
>> label_names =
\'airplane\'
\'automobile\'
\'bird\'
\'cat\'
\'deer\'
\'dog\'
\'frog\'
\'horse\'
\'ship\'
\'truck\'
>> categorical(integer_labels, 0:9, label_names)
3. categorical ⇒ numeric
c = categorical({\'Male\',\'Female\',\'Female\',\'Male\',\'Female\'});
n = grp2idx(c);