mxnet 神经网络训练和预测

https://mxnet.incubator.apache.org/tutorials/basic/module.html

import logging
import random
logging.getLogger().setLevel(logging.INFO)
 
import mxnet as mx
import numpy as np
 
mx.random.seed(1234)
np.random.seed(1234)
random.seed(1234)
 
# 准备数据
fname = mx.test_utils.download('https://s3.us-east-2.amazonaws.com/mxnet-public/letter_recognition/letter-recognition.data')
data = np.genfromtxt(fname=fname,delimiter=',')[:,1:]
label = np.array([ord(l.split(',')[0])-ord('A') for l in open(fname, 'r')])
 
batch_size = 32
ntrain = int(data.shape[0]*0.8)
 
train_iter = mx.io.NDArrayIter(data[:ntrain,:],label[:ntrain],batch_size,shuffle=True)
val_iter = mx.io.NDArrayIter(data[ntrain:,:],label[ntrain:],batch_size)
 
# 定义网络
net = mx.sym.Variable('data')
net = mx.sym.FullyConnected(net, name='fc1', num_hidden=64)
net = mx.sym.Activation(net, name='relu1', act_type="relu")
net = mx.sym.FullyConnected(net, name='fc2', num_hidden=26)
net = mx.sym.SoftmaxOutput(net, name='softmax')
mx.viz.plot_network(net, node_attrs={"shape":"oval","fixedsize":"false"})
 
# # 创建模块
mod = mx.mod.Module(symbol=net,
                    context=mx.cpu(),
                    data_names=['data'],
                    label_names=['softmax_label'])
 
# # 中层接口
# # 训练模型
# mod.bind(data_shapes=train_iter.provide_data,label_shapes=train_iter.provide_label)
# mod.init_params(initializer=mx.init.Uniform(scale=.1))
# mod.init_optimizer(optimizer='sgd',optimizer_params=(('learning_rate',0.1),))
# metric = mx.metric.create('acc')
#
# for epoch in range(100):
#     train_iter.reset()
#     metric.reset()
#     for batch in train_iter:
#         mod.forward(batch,is_train=True)
#         mod.update_metric(metric,batch.label)
#         mod.backward()
#         mod.update()
#     print('Epoch %d,Training %s' % (epoch,metric.get()))
 
# fit 高层接口
train_iter.reset()
mod = mx.mod.Module(symbol=net,
                    context=mx.cpu(),
                    data_names=['data'],
                    label_names=['softmax_label'])
 
mod.fit(train_iter,
        eval_data=val_iter,
        optimizer='sgd',
        optimizer_params={'learning_rate':0.1},
        eval_metric='acc',
        num_epoch=10)
 
# 预测和评估
y = mod.predict(val_iter)
assert y.shape == (4000,26)
 
# 评分
score = mod.score(val_iter,['acc'])
print("Accuracy score is %f"%(score[0][1]))
assert score[0][1] > 0.76, "Achieved accuracy (%f) is less than expected (0.76)" % score[0][1]
 
# 保存和加载
# 构造一个回调函数保存检查点
model_prefix = 'mx_mlp'
checkpoint = mx.callback.do_checkpoint(model_prefix)
 
mod = mx.mod.Module(symbol=net)
mod.fit(train_iter,num_epoch=5,epoch_end_callback=checkpoint)
 
sym, arg_params, aux_params = mx.model.load_checkpoint(model_prefix, 3)
assert sym.tojson() == net.tojson()
 
# assign the loaded parameters to the module
mod.set_params(arg_params, aux_params)
 
mod = mx.mod.Module(symbol=sym)
mod.fit(train_iter,
        num_epoch=21,
        arg_params=arg_params,
        aux_params=aux_params,
        begin_epoch=3)
assert score[0][1] > 0.77, "Achieved accuracy (%f) is less than expected (0.77)" % score[0][1]