深度信念网络回归算法Python程序.docx

需积分: 50 128 浏览量更新于2023-05-12 2 收藏 30KB DOCX 举报

身份认证购VIP最低享 7 折!

领优惠券(最高得80元）

资源详情

资源推荐

mian:

import numpy

import theano

import sys

import timeit

import csv

import math

import pickle

from DBNR import DBNR

from time import strftime

from sklearn.cross_validation import KFold

from sklearn.metrics import r2_score, mean_squared_error

def load_data(datasets=None, train_index=None, test_index=None):

x = datasets[0]

y = datasets[1]

with open(x, 'r') as g:

read = csv.reader(g)

gen = numpy.array(list(read), dtype='float64')

g.close()

with open(y, 'r') as f:

read = csv.reader(f)

fen = numpy.array(list(read), dtype='float64').flatten()

f.close()

max_ = max(fen)

min_ = min(fen)

fen = (fen - min_) / (max_ - min_)

''' x_train, x_test, y_train, y_test = train_test_split(gen, fen, test_size=0.1, random_state=42)'''

x_train, x_test = gen[train_index], gen[test_index]

y_train, y_test = fen[train_index], fen[test_index]

train_set = x_train, y_train

test_set = x_test, y_test

validation_set = x_test, y_test

def shared_dataset(data_xy):

data_x, data_y = data_xy

shared_x = theano.shared(numpy.asarray(data_x, dtype=theano.config.floatX))

shared_y = theano.shared(numpy.asarray(data_y, dtype=theano.config.floatX))

return shared_x, shared_y

train_set_x, train_set_y = shared_dataset(train_set)

val_set_x, val_set_y = shared_dataset(validation_set)

test_set_x, test_set_y = shared_dataset(test_set)

dataset = [(train_set_x, train_set_y), (val_set_x, val_set_y), (test_set_x, test_set_y)]

return dataset, max_

def test_DBNR(fold=None, finetune_lr=0.001, pretraining_epochs=200, pretrain_lr=0.001, k=1,

training_epochs=200, datasets=None, batch_size=1, n_ins=None,

hidden_layers_sizes=None, n_outs=None, train_index=None,

test_index=None, out_save=None, model_save=None):

dataset, max_y = load_data(datasets, train_index=train_index,

test_index=test_index)

train_set_x, train_set_y = dataset[0]

n_train_batches = train_set_x.get_value(borrow=True).shape[0] / batch_size

numpy_rng = numpy.random.RandomState(123)

print ('... membangun model')

dbnr = DBNR(numpy_rng=numpy_rng, n_ins=n_ins,

hidden_layers_size=hidden_layers_sizes, n_outs=n_outs)

###################################

''' PRETRAINING THE MODEL '''

###################################

print ('... setting pretraining function')

pretraining_fns = dbnr.pretraining_function(train_set_x=train_set_x, batch_size=batch_size, k=k)

print ('... pre-training model')

start_time = timeit.default_timer()

for i in range(dbnr.n_layers):

for epoch in range(pretraining_epochs):

c = []

for batch_index in range(n_train_batches):

c.append(pretraining_fns[i](index=batch_index, lr=pretrain_lr))

if epoch % 100 == 0:

print ('Pre-training layer ke-%i epoch ke-%d, rata-rata cost = ' % (i, epoch))

print (numpy.mean(c))

end_time =timeit.default_timer()

print >> sys.stderr, ('Pretraining selesai dalam waktu %.2fm' % ((end_time - start_time) / 60))

###################################

''' FINETUNING THE MODEL '''

###################################

print ('... setting finetuning functions')

train_fn, validate_model, test_model = dbnr.build_finetune_functions(

datasets=dataset, batch_size=batch_size, learning_rate=finetune_lr)

print ('... finetuning model')

patience = 4 * n_train_batches

patience_increase = 2.

improvement_threshold = 0.995

validation_frequency = min(n_train_batches, patience / 2)

best_validation_loss = numpy.inf

test_score = 0.

start_time = timeit.default_timer()

done_looping = False

epoch = 0

while (epoch < training_epochs): # and (not done_looping):

epoch += 1

for minibatch_index in range(n_train_batches):

minibatch_avg_cost = train_fn(minibatch_index)

iter = (epoch - 1) * n_train_batches + minibatch_index

if (iter + 1) % validation_frequency == 0:

validation_losses = validate_model()

this_validation_loss = numpy.mean(validation_losses)

print('epoch ke-%i, minibatch %i/%i, validation MSE sebesar %f '

% (epoch, minibatch_index + 1, n_train_batches,

this_validation_loss))

if this_validation_loss < best_validation_loss:

if (this_validation_loss < best_validation_loss * improvement_threshold):

patience = max(patience, iter * patience_increase)

best_validation_loss = this_validation_loss

best_iter = iter

test_losses = test_model()

test_score = numpy.mean(test_losses)

print((' epoch ke-%i, minibatch %i/%i, test MSE dari '

'best model adalah = %f ') %

(epoch, minibatch_index + 1, n_train_batches, test_score))

with open(model_save, 'wb') as f:

pickle.dump(dbnr, f)

if patience <= iter:

done_looping = True

break

end_time =timeit.default_timer()

print(('Optimalisasi selesai dengan score validasi terbaik = %f , '

'didapat pada iterasi ke- %i, '

'dengan performa test %f ') % (best_validation_loss, best_iter + 1, test_score) )

print >> sys.stderr, ('Finetuning selesai dalam waktu %.2fm' % ((end_time - start_time) / 60.))

predict(dataset, out_save, model_save, fold, max_y)

def predict(data=None, out_saved=None, model_saved=None, fold=None, y_max=None):

# load the saved model

classifier = pickle.load(open(model_saved))

# compile a predictor function

predict_model = theano.function(inputs=[classifier.x], outputs=classifier.y_pred)

# We can test it on some examples from test test

test_set_x, test_set_y = data[2]

test_set_x = test_set_x.get_value()

predicted_values = predict_model(test_set_x)

y_model = []

for i in range(len(predicted_values)):

y_model.append(predicted_values[i][0])

y_model = numpy.array(y_model) * y_max

numpy.set_printoptions(precision=4)

true_value = test_set_y.get_value() * y_max

r_squared = r2_score(true_value, y_model)

mean_a = numpy.mean(true_value)

mean_b = numpy.mean(y_model)

pears = (sum((true_value - mean_a) * (y_model - mean_b))) / math.sqrt((sum((true_value - mean_a) ** 2)) *

(sum((y_model - mean_b) ** 2)))

#r_square_adj = 1 - (((1 - r_squared) * (9 - 1)) / (9 - 1072 - 1))

mse = mean_squared_error(true_value, y_model)

剩余12页未读，继续阅读

紫薇大帝

粉丝: 34
资源: 8

会员权益专享

深度信念网络回归算法Python程序.docx

dbn-py.rar_DBN Python_DBN python实现_DBN python算法_dbn_python实现dbn

深度学习--深度信念网络(Deep-Belief-Network)

基于Python的深度信念网络程序

python mraptor.py file.docx

使用python将xx文件夹下的.sv文件复制到mode.docx文件并另存为.sv文件名的.docx文件

两个docx文档，1.docx和2.docx，需要将1.docx文档中的页眉页脚复制到2.docx，用java编写，使用开源jar包，比如poi，poi-tl，docx4j等等，请注意验证你提供的代码中的方法，麻烦给我源码。

.docx文件在vscode打开后。.docx文件发生了错误

python怎么打开.docx文件

将每个子文件夹里的.docx文件都合成一个.docx文件

将xx文件夹下的每个子文件夹里的.docx文件都合成一个.docx文件

请用python把word1.docx中指定位置的内容无格式填充到word2.docx

raise PackageNotFoundError( docx.opc.exceptions.PackageNotFoundError: Package not found at 'example.docx'

使用python编写代码将一个文件夹下的所有子文件夹下的.sv文件按照文件名称为model.docx的文件格式转化为与.sv同名的.docx的文件，页眉写上该.SV文件的名称

[('D:\\python项目\\分类分级\\test_dir\\1.docx',)]这种结构怎么取出里面的文本信息

用python编写一个小程序，实现需求： 1、只输出模板.docx的文本框的内容

python 将text文件如何转换为.docx文件

两个docx文档，1.docx和2.docx，需要将1.docx文档中的页眉页脚同步到2.docx，用java编写，使用开源jar包，比如poi，poi-tl，docx4j等等，麻烦给我源码

写一个提取.docx文档首行内容并以此重命名该文档程序

写一个提取.docx文档首行内容并以此重命名该文档的程序

会员权益专享

最新资源