神经网络实战记录2—neuron2单个神经元在逻辑斯蒂回归模型上的实现

最新推荐文章于 2025-01-07 02:11:27 发布
最新推荐文章于 2025-01-07 02:11:27 发布
阅读量540 收藏
点赞数
版权声明：本文为博主原创文章，遵循 CC 4.0 BY-SA 版权协议，转载请附上原文出处链接和本声明。
本文链接：https://blog.youkuaiyun.com/yuanjiu4221/article/details/100051849
版权
本文详细介绍了使用TensorFlow框架对CIFAR-10数据集进行预处理、构建神经网络并训练的过程。从数据加载、批处理到模型定义、损失函数计算及准确率评估均有涉及。

摘要生成于 C知道 ，由 DeepSeek-R1 满血版支持， 前往体验 >

1.打开文件夹
2.导入数据
3.处理数据
4.定义神经网络：
y
x
w
b
激活函数
loss
准确率
训练停止条件
5.训练神经网络：
定义：
batch_size
train_steps
test_steps
开始训练
6.其他杂记：
（1）
出现这个错误是提醒要更新tensorflow的版本了，anaconda里base(root)—not installed-tensorlow-gpu—apply即可
。。。但是好像我更新了还是抱这个警告，先不管了放在这
（2）直接运行第四个框的时候会报一个“loss未定义”的错，这时候把第三个框再运行一下就解决了
（3）话不多说，最终部分代码和结果如下：
Python 3
File
Edit
View
Insert
Cell
Kernel
Widgets
Help

./…/…/cifar-10-batches-py
1
import tensorflow as tf
2
import os
3
import pickle
4
import numpy as np
5
​
6
CIFAR_DIR = “F:\lijingwen\Muke\muke\cifar-10-batches-py”
7
print(os.listdir(CIFAR_DIR))
[‘batches.meta’, ‘data_batch_1’, ‘data_batch_2’, ‘data_batch_3’, ‘data_batch_4’, ‘data_batch_5’, ‘readme.html’, ‘test_batch’]
1
def load_data(filename):
2
“”“read data from data file.”""
3
with open(filename, ‘rb’) as f:
4
data = pickle.load(f, encoding=‘bytes’)
5
return data[b’data’], data[b’labels’]
6
​
7

tensorflow.Dataset.

8
class CifarData:
9
def init(self, filenames, need_shuffle):
10
all_data = []
11
all_labels = []
12
for filename in filenames:
13
data, labels = load_data(filename)
14
for item, label in zip(data, labels):
15
if label in [0, 1]:
16
all_data.append(item)
17
all_labels.append(label)
18
self._data = np.vstack(all_data)
19
self._data = self._data / 127.5 - 1
20
self._labels = np.hstack(all_labels)
21
print(self._data.shape)
22
print(self._labels.shape)
23

24
self._num_examples = self._data.shape[0]
25
self._need_shuffle = need_shuffle
26
self._indicator = 0
27
if self._need_shuffle:
28
self._shuffle_data()
29

30
def _shuffle_data(self):
31
# [0,1,2,3,4,5] -> [5,3,2,4,0,1]
32
p = np.random.permutation(self._num_examples)
33
self._data = self._data[p]
34
self._labels = self._labels[p]
35

36
def next_batch(self, batch_size):
37
“”“return batch_size examples as a batch.”""
38
end_indicator = self._indicator + batch_size
39
if end_indicator > self._num_examples:
40
if self._need_shuffle:
41
self._shuffle_data()
42
self._indicator = 0
43
end_indicator = batch_size
44
else:
45
raise Exception(“have no more examples”)
46
if end_indicator > self._num_examples:
47
raise Exception(“batch size is larger than all examples”)
48
batch_data = self._data[self._indicator: end_indicator]
49
batch_labels = self._labels[self._indicator: end_indicator]
50
self.indicator = end_indicator
51
return batch_data, batch_labels
52
​
53
train_filenames = [os.path.join(CIFAR_DIR, 'data_batch%d’ % i) for i in range(1, 6)]
54
test_filenames = [os.path.join(CIFAR_DIR, ‘test_batch’)]
55
​
56
train_data = CifarData(train_filenames, True)
57
test_data = CifarData(test_filenames, False)
(10000, 3072)
(10000,)
(2000, 3072)
(2000,)
1
x = tf.placeholder(tf.float32, [None, 3072])
2

[None]

3
y = tf.placeholder(tf.int64, [None])
4
​
5

(3072, 1)

6
w = tf.get_variable(‘w’, [x.get_shape()[-1], 1],
7
initializer=tf.random_normal_initializer(0, 1))
8

(1, )

9
b = tf.get_variable(‘b’, [1],
10
initializer=tf.constant_initializer(0.0))
11
​
12

[None, 3072] * [3072, 1] = [None, 1]

13
y_ = tf.matmul(x, w) + b
14
​
15

[None, 1]

16
p_y_1 = tf.nn.sigmoid(y_)
17

[None, 1]

18
y_reshaped = tf.reshape(y, (-1, 1))
19
y_reshaped_float = tf.cast(y_reshaped, tf.float32)
20
loss = tf.reduce_mean(tf.square(y_reshaped_float - p_y_1))
21
​
22

bool

23
predict = p_y_1 > 0.5
24

[1,0,1,1,1,0,0,0]

25
correct_prediction = tf.equal(tf.cast(predict, tf.int64), y_reshaped)
26
accuracy = tf.reduce_mean(tf.cast(correct_prediction, tf.float64))
27
​
28
with tf.name_scope(‘train_op’):
29
train_op = tf.train.AdamOptimizer(1e-3).minimize(loss)
WARNING:tensorflow:From F:\Anaconda3\lib\site-packages\tensorflow\python\framework\op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version.
Instructions for updating:
Colocations handled automatically by placer.
1
init = tf.global_variables_initializer()
2
batch_size = 20
3
train_steps = 100000
4
test_steps = 100
5
​
6
with tf.Session() as sess:
7
sess.run(init)
8
for i in range(train_steps):
9
batch_data, batch_labels = train_data.next_batch(batch_size)
10
loss_val, acc_val, _ = sess.run(
11
[loss, accuracy, train_op],
12
feed_dict={
13
x: batch_data,
14
y: batch_labels})
15
if (i+1) % 500 == 0:
16
print(’[Train] Step: %d, loss: %4.5f, acc: %4.5f’ % (i+1, loss_val, acc_val))
17
if (i+1) % 5000 == 0:
18
test_data = CifarData(test_filenames, False)
19
all_test_acc_val = []
20
for j in range(test_steps):
21
test_batch_data, test_batch_labels
22
= test_data.next_batch(batch_size)
23
test_acc_val = sess.run(
24
[accuracy],
25
feed_dict = {
26
x: test_batch_data,
27
y: test_batch_labels
28
})
29
all_test_acc_val.append(test_acc_val)
30
test_acc = np.mean(all_test_acc_val)
31
print(’[Test ] Step: %d, acc: %4.5f’ % (i+1, test_acc))
32

33

34

[Train] Step: 500, loss: 0.33741, acc: 0.65000
[Train] Step: 1000, loss: 0.29905, acc: 0.70000
[Train] Step: 1500, loss: 0.15000, acc: 0.85000
[Train] Step: 2000, loss: 0.10277, acc: 0.90000
[Train] Step: 2500, loss: 0.10000, acc: 0.90000
[Train] Step: 3000, loss: 0.19999, acc: 0.80000
[Train] Step: 3500, loss: 0.09566, acc: 0.90000
[Train] Step: 4000, loss: 0.31656, acc: 0.65000
[Train] Step: 4500, loss: 0.24712, acc: 0.75000
[Train] Step: 5000, loss: 0.26500, acc: 0.70000
(2000, 3072)
(2000,)
[Test ] Step: 5000, acc: 0.80150
[Train] Step: 5500, loss: 0.20336, acc: 0.80000
[Train] Step: 6000, loss: 0.05478, acc: 0.95000
[Train] Step: 6500, loss: 0.12125, acc: 0.85000
[Train] Step: 7000, loss: 0.24541, acc: 0.75000
[Train] Step: 7500, loss: 0.25299, acc: 0.75000
[Train] Step: 8000, loss: 0.10263, acc: 0.90000
[Train] Step: 8500, loss: 0.25992, acc: 0.70000
[Train] Step: 9000, loss: 0.10083, acc: 0.90000
[Train] Step: 9500, loss: 0.10000, acc: 0.90000
[Train] Step: 10000, loss: 0.26703, acc: 0.70000
(2000, 3072)
(2000,)
[Test ] Step: 10000, acc: 0.81400
[Train] Step: 10500, loss: 0.10108, acc: 0.90000
[Train] Step: 11000, loss: 0.24926, acc: 0.75000
[Train] Step: 11500, loss: 0.27655, acc: 0.65000
[Train] Step: 12000, loss: 0.22258, acc: 0.75000
[Train] Step: 12500, loss: 0.05000, acc: 0.95000
[Train] Step: 13000, loss: 0.10006, acc: 0.90000
[Train] Step: 13500, loss: 0.14959, acc: 0.85000
[Train] Step: 14000, loss: 0.32116, acc: 0.65000
[Train] Step: 14500, loss: 0.08510, acc: 0.90000
[Train] Step: 15000, loss: 0.25159, acc: 0.75000
(2000, 3072)
(2000,)
[Test ] Step: 15000, acc: 0.81800
[Train] Step: 15500, loss: 0.05233, acc: 0.95000
[Train] Step: 16000, loss: 0.17504, acc: 0.80000
[Train] Step: 16500, loss: 0.10001, acc: 0.90000
[Train] Step: 17000, loss: 0.14993, acc: 0.85000
[Train] Step: 17500, loss: 0.19063, acc: 0.80000
[Train] Step: 18000, loss: 0.10990, acc: 0.90000
[Train] Step: 18500, loss: 0.00076, acc: 1.00000
[Train] Step: 19000, loss: 0.17156, acc: 0.80000
[Train] Step: 19500, loss: 0.20010, acc: 0.80000
[Train] Step: 20000, loss: 0.30069, acc: 0.70000
(2000, 3072)
(2000,)
[Test ] Step: 20000, acc: 0.81600
[Train] Step: 20500, loss: 0.15001, acc: 0.85000
[Train] Step: 21000, loss: 0.10002, acc: 0.90000
[Train] Step: 21500, loss: 0.15820, acc: 0.85000
[Train] Step: 22000, loss: 0.05046, acc: 0.95000
[Train] Step: 22500, loss: 0.10248, acc: 0.90000
[Train] Step: 23000, loss: 0.05405, acc: 0.95000
[Train] Step: 23500, loss: 0.30000, acc: 0.70000
[Train] Step: 24000, loss: 0.24984, acc: 0.75000
[Train] Step: 24500, loss: 0.10027, acc: 0.90000
[Train] Step: 25000, loss: 0.00546, acc: 1.00000
(2000, 3072)
(2000,)
[Test ] Step: 25000, acc: 0.81900
[Train] Step: 25500, loss: 0.15296, acc: 0.85000
[Train] Step: 26000, loss: 0.10159, acc: 0.90000
[Train] Step: 26500, loss: 0.20073, acc: 0.80000
[Train] Step: 27000, loss: 0.18369, acc: 0.80000
[Train] Step: 27500, loss: 0.21014, acc: 0.80000
[Train] Step: 28000, loss: 0.30000, acc: 0.70000
[Train] Step: 28500, loss: 0.09921, acc: 0.90000
[Train] Step: 29000, loss: 0.10718, acc: 0.90000
[Train] Step: 29500, loss: 0.20979, acc: 0.80000
[Train] Step: 30000, loss: 0.34932, acc: 0.65000
(2000, 3072)
(2000,)
[Test ] Step: 30000, acc: 0.82100
[Train] Step: 30500, loss: 0.25288, acc: 0.75000
[Train] Step: 31000, loss: 0.12723, acc: 0.85000
[Train] Step: 31500, loss: 0.15197, acc: 0.85000
[Train] Step: 32000, loss: 0.10018, acc: 0.90000
[Train] Step: 32500, loss: 0.20464, acc: 0.75000
[Train] Step: 33000, loss: 0.10629, acc: 0.90000
[Train] Step: 33500, loss: 0.25627, acc: 0.75000
[Train] Step: 34000, loss: 0.29362, acc: 0.70000
[Train] Step: 34500, loss: 0.15014, acc: 0.85000
[Train] Step: 35000, loss: 0.00004, acc: 1.00000
(2000, 3072)
(2000,)
[Test ] Step: 35000, acc: 0.81900
[Train] Step: 35500, loss: 0.24982, acc: 0.75000
[Train] Step: 36000, loss: 0.20046, acc: 0.80000
[Train] Step: 36500, loss: 0.15197, acc: 0.85000
[Train] Step: 37000, loss: 0.05000, acc: 0.95000
[Train] Step: 37500, loss: 0.20000, acc: 0.80000
[Train] Step: 38000, loss: 0.12759, acc: 0.85000
[Train] Step: 38500, loss: 0.25005, acc: 0.75000
[Train] Step: 39000, loss: 0.05000, acc: 0.95000
[Train] Step: 39500, loss: 0.05000, acc: 0.95000
[Train] Step: 40000, loss: 0.19657, acc: 0.80000
(2000, 3072)
(2000,)
[Test ] Step: 40000, acc: 0.82000
[Train] Step: 40500, loss: 0.10003, acc: 0.90000
[Train] Step: 41000, loss: 0.20007, acc: 0.80000
[Train] Step: 41500, loss: 0.25698, acc: 0.75000
[Train] Step: 42000, loss: 0.15053, acc: 0.85000
[Train] Step: 42500, loss: 0.25112, acc: 0.75000
[Train] Step: 43000, loss: 0.20570, acc: 0.80000
[Train] Step: 43500, loss: 0.17739, acc: 0.80000
[Train] Step: 44000, loss: 0.15514, acc: 0.85000
[Train] Step: 44500, loss: 0.05007, acc: 0.95000
[Train] Step: 45000, loss: 0.14999, acc: 0.85000
(2000, 3072)
(2000,)
[Test ] Step: 45000, acc: 0.82150
[Train] Step: 45500, loss: 0.20068, acc: 0.80000
[Train] Step: 46000, loss: 0.10000, acc: 0.90000
[Train] Step: 46500, loss: 0.10761, acc: 0.90000
[Train] Step: 47000, loss: 0.15000, acc: 0.85000
[Train] Step: 47500, loss: 0.20128, acc: 0.80000
[Train] Step: 48000, loss: 0.15010, acc: 0.85000
[Train] Step: 48500, loss: 0.10000, acc: 0.90000
[Train] Step: 49000, loss: 0.14994, acc: 0.85000
[Train] Step: 49500, loss: 0.10000, acc: 0.90000
[Train] Step: 50000, loss: 0.20641, acc: 0.80000
(2000, 3072)
(2000,)
[Test ] Step: 50000, acc: 0.81600
[Train] Step: 50500, loss: 0.15067, acc: 0.85000
[Train] Step: 51000, loss: 0.29985, acc: 0.70000
[Train] Step: 51500, loss: 0.09935, acc: 0.90000
[Train] Step: 52000, loss: 0.19986, acc: 0.80000
[Train] Step: 52500, loss: 0.10071, acc: 0.90000
[Train] Step: 53000, loss: 0.10000, acc: 0.90000
[Train] Step: 53500, loss: 0.10568, acc: 0.90000
[Train] Step: 54000, loss: 0.15523, acc: 0.85000
[Train] Step: 54500, loss: 0.20090, acc: 0.80000
[Train] Step: 55000, loss: 0.15253, acc: 0.85000
(2000, 3072)
(2000,)
[Test ] Step: 55000, acc: 0.82000
[Train] Step: 55500, loss: 0.27071, acc: 0.70000
[Train] Step: 56000, loss: 0.24246, acc: 0.75000
[Train] Step: 56500, loss: 0.05011, acc: 0.95000
[Train] Step: 57000, loss: 0.18971, acc: 0.80000
[Train] Step: 57500, loss: 0.05087, acc: 0.95000
[Train] Step: 58000, loss: 0.22394, acc: 0.80000
[Train] Step: 58500, loss: 0.24600, acc: 0.75000
[Train] Step: 59000, loss: 0.14219, acc: 0.85000
[Train] Step: 59500, loss: 0.19999, acc: 0.80000
[Train] Step: 60000, loss: 0.20027, acc: 0.80000
(2000, 3072)
(2000,)
[Test ] Step: 60000, acc: 0.81700
[Train] Step: 60500, loss: 0.10153, acc: 0.90000
[Train] Step: 61000, loss: 0.20000, acc: 0.80000
[Train] Step: 61500, loss: 0.09850, acc: 0.90000
[Train] Step: 62000, loss: 0.19987, acc: 0.80000
[Train] Step: 62500, loss: 0.09999, acc: 0.90000
[Train] Step: 63000, loss: 0.05012, acc: 0.95000
[Train] Step: 63500, loss: 0.10001, acc: 0.90000
[Train] Step: 64000, loss: 0.05155, acc: 0.95000
[Train] Step: 64500, loss: 0.05357, acc: 0.95000
[Train] Step: 65000, loss: 0.05748, acc: 0.95000
(2000, 3072)
(2000,)
[Test ] Step: 65000, acc: 0.81800
[Train] Step: 65500, loss: 0.10019, acc: 0.90000
[Train] Step: 66000, loss: 0.06018, acc: 0.95000
[Train] Step: 66500, loss: 0.19917, acc: 0.80000
[Train] Step: 67000, loss: 0.30000, acc: 0.70000
[Train] Step: 67500, loss: 0.20122, acc: 0.80000
[Train] Step: 68000, loss: 0.15687, acc: 0.85000
[Train] Step: 68500, loss: 0.14914, acc: 0.85000
[Train] Step: 69000, loss: 0.00008, acc: 1.00000
[Train] Step: 69500, loss: 0.15009, acc: 0.85000
[Train] Step: 70000, loss: 0.15000, acc: 0.85000
(2000, 3072)
(2000,)
[Test ] Step: 70000, acc: 0.81900
[Train] Step: 70500, loss: 0.20019, acc: 0.80000
[Train] Step: 71000, loss: 0.10350, acc: 0.90000
[Train] Step: 71500, loss: 0.10000, acc: 0.90000
[Train] Step: 72000, loss: 0.16096, acc: 0.85000
[Train] Step: 72500, loss: 0.16597, acc: 0.85000
[Train] Step: 73000, loss: 0.14999, acc: 0.85000
[Train] Step: 73500, loss: 0.10000, acc: 0.90000
[Train] Step: 74000, loss: 0.10053, acc: 0.90000
[Train] Step: 74500, loss: 0.10057, acc: 0.90000
[Train] Step: 75000, loss: 0.19972, acc: 0.80000
(2000, 3072)
(2000,)
[Test ] Step: 75000, acc: 0.81900
[Train] Step: 75500, loss: 0.05002, acc: 0.95000
[Train] Step: 76000, loss: 0.05000, acc: 0.95000
[Train] Step: 76500, loss: 0.15290, acc: 0.85000
[Train] Step: 77000, loss: 0.10020, acc: 0.90000
[Train] Step: 77500, loss: 0.20000, acc: 0.80000
[Train] Step: 78000, loss: 0.19840, acc: 0.80000
[Train] Step: 78500, loss: 0.24995, acc: 0.75000
[Train] Step: 79000, loss: 0.20000, acc: 0.80000
[Train] Step: 79500, loss: 0.00000, acc: 1.00000
[Train] Step: 80000, loss: 0.05130, acc: 0.95000
(2000, 3072)
(2000,)
[Test ] Step: 80000, acc: 0.81850
[Train] Step: 80500, loss: 0.20014, acc: 0.80000
[Train] Step: 81000, loss: 0.25691, acc: 0.75000
[Train] Step: 81500, loss: 0.10194, acc: 0.90000
[Train] Step: 82000, loss: 0.15047, acc: 0.85000
[Train] Step: 82500, loss: 0.25760, acc: 0.75000
[Train] Step: 83000, loss: 0.20254, acc: 0.80000
[Train] Step: 83500, loss: 0.10000, acc: 0.90000
[Train] Step: 84000, loss: 0.20021, acc: 0.80000
[Train] Step: 84500, loss: 0.25000, acc: 0.75000
[Train] Step: 85000, loss: 0.19997, acc: 0.80000
(2000, 3072)
(2000,)
[Test ] Step: 85000, acc: 0.81950
[Train] Step: 85500, loss: 0.15016, acc: 0.85000
[Train] Step: 86000, loss: 0.35113, acc: 0.65000
[Train] Step: 86500, loss: 0.20030, acc: 0.80000
[Train] Step: 87000, loss: 0.15027, acc: 0.85000
[Train] Step: 87500, loss: 0.15078, acc: 0.85000
[Train] Step: 88000, loss: 0.10000, acc: 0.90000
[Train] Step: 88500, loss: 0.05000, acc: 0.95000
[Train] Step: 89000, loss: 0.20001, acc: 0.80000
[Train] Step: 89500, loss: 0.10005, acc: 0.90000
[Train] Step: 90000, loss: 0.10808, acc: 0.90000
(2000, 3072)
(2000,)
[Test ] Step: 90000, acc: 0.82200
[Train] Step: 90500, loss: 0.20000, acc: 0.80000
[Train] Step: 91000, loss: 0.20201, acc: 0.80000
[Train] Step: 91500, loss: 0.10000, acc: 0.90000
[Train] Step: 92000, loss: 0.05425, acc: 0.95000
[Train] Step: 92500, loss: 0.20619, acc: 0.80000
[Train] Step: 93000, loss: 0.09443, acc: 0.90000
[Train] Step: 93500, loss: 0.10001, acc: 0.90000
[Train] Step: 94000, loss: 0.19999, acc: 0.80000
[Train] Step: 94500, loss: 0.20418, acc: 0.80000
[Train] Step: 95000, loss: 0.27056, acc: 0.70000
(2000, 3072)
(2000,)
[Test ] Step: 95000, acc: 0.81800
[Train] Step: 95500, loss: 0.20252, acc: 0.80000
[Train] Step: 96000, loss: 0.19851, acc: 0.80000
[Train] Step: 96500, loss: 0.20412, acc: 0.80000
[Train] Step: 97000, loss: 0.10000, acc: 0.90000
[Train] Step: 97500, loss: 0.15101, acc: 0.85000
[Train] Step: 98000, loss: 0.00317, acc: 1.00000
[Train] Step: 98500, loss: 0.10255, acc: 0.90000
[Train] Step: 99000, loss: 0.15000, acc: 0.85000
[Train] Step: 99500, loss: 0.15000, acc: 0.85000
[Train] Step: 100000, loss: 0.15025, acc: 0.85000
(2000, 3072)
(2000,)
[Test ] Step: 100000, acc: 0.81850
1
​

SSssss630
关注
机器学习实验四：深度学习图像生成（Part two：DeepDream）
蓝多多的小仓库
10-23 2771
关于这个警告：Instructions for updating: Colocations handled automatically by placer.不必理会。网上给了一种解释： 其它的没啥，文章设置成转载了，代码是实验指导书上的，指导书来源姜老师，仅供学习存档。（代码修改参照Part one即可） 模型文件及图像下载链接（无需积分，直接获取下载码即可免费下载，如遇404说明资源在审核中）： https://download.csdn.net/download/qq_43554335/340
Hindmarsh-Rose神经元模型matlab仿真程序.txt
09-27
自己写的HR神经元模型，亲测可用，勿担心，其中I_ext为外加刺激电流，改变此值可得到不同效果
深度学习 神经网络 逻辑斯蒂回归模型
Jason
04-25 701
神经元二分类多分类目标函数平方差损失交叉熵损失梯度下降 神经元 hW,b(x)=f(WTx)=f(∑i=13Wixi+b)h_{W, b}(x) = f(W^Tx) = f(\sum^{3}_{i=1}W_ix_i+b)hW,b​(x)=f(WTx)=f(i=1∑3​Wi​xi​+b) WWW 表示权重 xxx 表示特征 f()f()f() 表示激活函数 bbb 是偏置, 如图中的 +1, ...
1.1 神经网络入门-逻辑斯底回归模型
weixin_34315485的博客
01-26 233
1.1 神经元-逻辑斯底回归模型 神经元 ：神经网络的最小结构，多个神经元组合在一起，就可以形成一个神经网络 逻辑回归模型：神经元经过设置之后就能变成一个模型，叫做逻辑斯底回归模型 神经网络训练：知道了神经网络的结构之后，如何调整参数，使得神经网络能够在数据上学到规律，做出好的预测 神经元的基本结构： ​ 注：b可以泛化为1*b，其中的1泛化为输入x，b泛化为w。 ...
《神经网络》历程
好学近乎知，知耻近乎勇
06-07 830
1、神经网络编程入门
逻辑斯蒂回归（Logistic Regression）
Allenlzcoder的博客
03-31 357
逻辑斯蒂回归（Logistic Regression） 博客转载自：https://blog.csdn.net/daunxx/article/details/51816588 一篇写逻辑斯蒂回归挺好的博客，改天转载过来~
【AI学习笔记3】神经元、神经网络与感知机 Neuron、Neural Network and Perceptron
最新发布
2301_76992325的博客
01-07 1033
或门在(x1, x2) = (0, 0) 时输出0，在(x1, x2) 为(0, 1)、(1, 0)、(1, 1) 时输出1。上图中，○表示0，△表示1。或门的情况下，当权重参数为(b,w_1,w_2)=(-0.5,1.0,1.0)时，可满足真值表条件。如果用感知机来表示与门，可以把（w1,w2,θ）分别设置为（1, 1, 1），或者设置为(0.5, 0.5, 0.8)，实际上有无穷多种取值都可以达成预期。这里，将最左边的一列称为第 0 层，中间的一列称为第 1 层，最右边的一列称为第 2 层。
SingleNeuronPIDController.rar_neuron matlab_pid_pid参数调整_单神经元_单神经
07-14
1. **神经网络结构**：定义单个神经元的模型，包括输入、输出和连接权重。 2. **PID参数映射**：如何将神经元的权重和偏置映射到PID控制器的P、I、D参数上。 3. **训练算法**：用于调整神经元参数的优化算法，例如...
基于BP神经网络与单神经元自适应神经网络的PID控制器
12-27
另一方面，单神经元自适应神经网络（Single Neuron Adaptive Control，简称SNAC）具有单个神经元结构，其适应性强、计算简单，适用于实时在线控制。SNAC通过自我学习和调整权重来实现系统的动态控制。结合BP神经网络...
matlab的欧拉方法代码-neuron-network-models:动态耦合激发大脑中神经元的模型以产生复杂的网络同步
05-26
在single_neuron_models目录中，运行着一些程序来模拟单个激发神经元的行为。每个程序的顶部都有一些示例运行。 OneNeuronTau.m：基于Tau常数的简单模型 OneNeuronIzhInF.m：伊兹凯维奇着名的“整合并发射”神经元...
published-neuron-models:博士神经元模型
04-04
它允许科学家们构建详细的单个神经元模型，甚至是大规模的神经网络模型，从而更好地理解神经系统的运作机制。通过NEURON，研究人员可以模拟电压门控离子通道的行为，观察动作电位的产生和传播，以及研究不同类型的...
python不能分配给操作员_python-Tensorflow无法分配设备进行操作
weixin_39732316的博客
12-15 513
我正在尝试在计算机上运行NVidia’s face generating demo.我正在使用Windows10.我已经下载了源代码,并试图按照页面下方的步骤进行操作.我已经为我的GTX1060安装了最新的NVidia驱动程序,该驱动程序应该是支持cuda功能的设备.我已经安装了Cuda Toolkit和TensorFlow所需的cuDNN SDK.但是,运行import_example.py脚本...
python画损失函数图_深度学习笔记48_你也可以成为梵高_损失函数算法实践
weixin_39929096的博客
12-08 558
from keras import backend as K# constant变量参数来定义target_image = K.constant(preprocess_image(target_image_path))style_reference_image = K.constant(preprocess_image(style_reference_image_path))# 用palcehol...
Jetson-Nano开箱配置及Tensorflow安装使用
热门推荐
dwSun的博客
04-02 5万+
Jetson-Nano开箱配置及Tensorflow安装使用 简介 最近老黄发布了新的硬件，号称Nvidia良心之作的99美元AI硬件Jetson-Nano: 该设备使用的是Maxwell（比pascal古老一点）架构的GPU，有128个Cuda核心，运算能力472G，看数值似乎一般般，但是考虑这是一个只有5W的嵌入式设备，这种运算能力已经相当可观了，足够部署一些小模型在终端设备上。 同时，Je...
深度学习之TensorFlow（五）学习率
qq_45416295的博客
02-29 348
文章目录学习率 学习率 学习率（learning_rate）：每次参数更新的幅度 wn+1=wn−learnin_grate∗∇w_{n+1}=w_n-learnin\_grate*\nablawn+1​=wn​−learnin_grate∗∇ wn+1w_{n_+1}wn+​1​是更新后的参数，wnw_nwn​是当前参数，∇\nabla∇是损失函数的梯度(导数) 例如损失函数loss=(w+1...
tensorflow学习 -- 变量管理与模型持久化
周永行的博客
05-20 926
1、变量管理 tensorflow的变量管理主要通过两个函数：tf.Variable和tf.get_varible来完成，前者用来创建一个变量，后者可以创建变量或者获取变量。在创建变量的时候，两者的功能是等价的。tf.Variable可以不传入name参数，但是tf.get_varible必须使用name参数以便用这个参数去创建或者获取变量。当使用tf.get_varible的时候，tf.get...
果蝇ACC运动神经元模型在XPP与Neuron软件中的实现
Neuron是一个广泛使用的神经元模拟器，用于建立和模拟复杂的神经网络和单个神经元的电生理特性。Neuron模拟器可以模拟离子通道、突触传递以及更复杂的神经元和突触网络的行为。在这个研究中，Neuron用于实现ACC运动...
评论
添加红包

请填写红包祝福语或标题

个

红包个数最小为10个

元

红包金额最低5元

当前余额3.43元 前往充值 >
需支付：10.00元
成就一亿技术人!
领取后你会自动成为博主和红包主的粉丝 规则
hope_wisdom
发出的红包
实付元
使用余额支付
点击重新获取
扫码支付
钱包余额 0

抵扣说明：

1.余额是钱包充值的虚拟货币，按照1:1的比例进行支付金额的抵扣。
2.余额无法直接购买下载，可以购买VIP、付费专栏及课程。

余额充值