多层感知机MLP的gluon版分类minist

最新推荐文章于 2024-10-16 20:53:10 发布

weixin_33725272

最新推荐文章于 2024-10-16 20:53:10 发布

阅读量149

点赞数

CC 4.0 BY-SA版权

文章标签：人工智能

原文链接：http://www.cnblogs.com/TreeDream/p/10021237.html

本文介绍了一个使用Gluon库实现的多层感知器（MLP）模型，该模型用于Fashion MNIST数据集上的图像分类任务。通过定义网络结构、初始化参数、设置损失函数和优化器，进行5轮训练，最终达到了较高的测试准确率。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

MLP_Gluon

In [2]:

import gluonbook as gb
from mxnet import gluon, init
from mxnet.gluon import loss as gloss,nn

In [4]:

net = nn.Sequential()
net.add(nn.Dense(256,activation='relu'),nn.Dense(10))
net.initialize(init.Normal(sigma=0.01))

In [5]:

batch_size = 256
train_iter, test_iter = gb.load_data_fashion_mnist(batch_size)

损失函数

In [6]:

loss = gloss.SoftmaxCrossEntropyLoss()
trainer = gluon.Trainer(net.collect_params(),'sgd',{'learning_rate':0.5})
num_epochs = 5
gb.train_ch3(net,train_iter,test_iter,loss,num_epochs,batch_size,None,None,trainer)

epoch 1, loss 0.8074, train acc 0.700, test acc 0.829
epoch 2, loss 0.4819, train acc 0.823, test acc 0.852
epoch 3, loss 0.4306, train acc 0.840, test acc 0.855
epoch 4, loss 0.3935, train acc 0.856, test acc 0.856
epoch 5, loss 0.3714, train acc 0.863, test acc 0.865

转载于:https://www.cnblogs.com/TreeDream/p/10021237.html