Ex_treme's blog.

神经元实现（二分类logistic回归模型实现）（2）

二分类 logistic
2018/11/13 Share
  • 调整了数据集的测试代码
  • 新增了训练的测试代码

数据集的测试代码

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
def load_data(filename):
    """read date from data file."""
    with open(filename,'rb') as f:
        data = pickle.load(f,encoding='bytes')
        return data[b'data'],data[b'labels']
    
class CifarData:
    def __init__(self,filenames,need_shuffle):
        all_data = []
        all_labels = []
        for filename in filenames:
            data,labels = load_data(filename)
            for item,label in zip(data,labels):
                if label in [0,1]:
                    all_data.append(item)
                    all_labels.append(label)
        self._data = np.vstack(all_data)
        self._data = self._data / 127.5 -1
        self._labels = np.hstack(all_labels)
        print(self._data.shape)
        print(self._labels.shape)
        
        self._num_examples = self._data.shape[0]
        self._need_shuffle = need_shuffle
        self._indicator = 0
        if self._need_shuffle:
            self._shuffle_data()
    
    def _shuffle_data(self):
        # [0,1,2,3,4,5] - > [5,3,2,4,0,1]
        p = np.random.permutation(self._num_examples)
        self._data = self._data[p]
        self._labels = self._labels[p]
        
    def next_batch(self, batch_size):
        """return batch_size examples as a batch."""
        end_indicator = self._indicator + batch_size
        if end_indicator > self._num_examples:
            if self._need_shuffle:
                self._shuffle_data()
                self._indicator = 0
                end_indicator = batch_size
            else:
                raise Exception("have no more examples")
        if end_indicator > self._num_examples:
            raise Exception("batch size is larger than all examples")
        batch_data = self._data[self._indicator:end_indicator]
        batch_labels = self._labels[self._indicator: end_indicator]
        self._indicator = end_indicator
        return batch_data, batch_labels

train_filenames = [os.path.join(CIFAR_DIR,'data_batch_%d' % i ) for i in range(1,6)]
test_filenames = [os.path.join(CIFAR_DIR,'test_batch')]

train_data = CifarData(train_filenames,True)
test_data = CifarData(test_filenames,False)

# batch_data,batch_labels = train_data.next_batch(10)
# print(batch_data)
# print(batch_labels)

(10000, 3072)

(10000,)

(2000, 3072)

(2000,)

新增了训练的测试代码

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
init = tf.global_variables_initializer()
batch_size = 20
train_steps = 100000
test_steps = 100

with tf.Session() as sess:
    sess.run(init)
    for i in range(train_steps):
        batch_data,batch_labels = train_data.next_batch(batch_size)
        loss_val, accu_val, _ = sess.run(
            [loss,accuracy,train_op],
            feed_dict={
                x: batch_data,
                y: batch_labels})
        if (i+1) % 500 == 0:
            print('[Train] Step: %d, loss:%4.5f, acc: %4.5f'
                 % (i+1,loss_val,accu_val))
        if (i+1) % 5000 == 0:
            test_data = CifarData(test_filenames,False)
            all_test_acc_val = []
            for j in range(test_steps):
                test_batch_data, test_batch_labels = test_data.next_batch(batch_size)
                test_acc_val = sess.run(
            [accuracy],
            feed_dict={
                x: test_batch_data,
                y: test_batch_labels})
                all_test_acc_val.append(test_acc_val)
            test_acc = np.mean(all_test_acc_val)
            print("[Test] Step: %d, acc: %4.5f" % (i+1,test_acc))

训练结果：

[Train] Step: 500, loss:0.34972, acc: 0.65000
[Train] Step: 1000, loss:0.25081, acc: 0.75000
[Train] Step: 1500, loss:0.32387, acc: 0.65000
[Train] Step: 2000, loss:0.19776, acc: 0.80000
[Train] Step: 2500, loss:0.15003, acc: 0.85000
[Train] Step: 3000, loss:0.13229, acc: 0.85000
[Train] Step: 3500, loss:0.20558, acc: 0.80000
[Train] Step: 4000, loss:0.29615, acc: 0.70000
[Train] Step: 4500, loss:0.08947, acc: 0.90000
[Train] Step: 5000, loss:0.24596, acc: 0.75000
(2000, 3072)
(2000,)
[Test] Step: 5000, acc: 0.79400
[Train] Step: 5500, loss:0.15080, acc: 0.85000
[Train] Step: 6000, loss:0.13784, acc: 0.85000
[Train] Step: 6500, loss:0.32210, acc: 0.65000
[Train] Step: 7000, loss:0.19997, acc: 0.80000
[Train] Step: 7500, loss:0.30646, acc: 0.65000
[Train] Step: 8000, loss:0.29746, acc: 0.70000
[Train] Step: 8500, loss:0.34132, acc: 0.65000
[Train] Step: 9000, loss:0.24997, acc: 0.75000
[Train] Step: 9500, loss:0.01327, acc: 1.00000
[Train] Step: 10000, loss:0.10051, acc: 0.90000
(2000, 3072)
(2000,)
[Test] Step: 10000, acc: 0.80700
[Train] Step: 10500, loss:0.15171, acc: 0.85000
[Train] Step: 11000, loss:0.16315, acc: 0.80000
[Train] Step: 11500, loss:0.03247, acc: 0.95000
[Train] Step: 12000, loss:0.11663, acc: 0.85000
[Train] Step: 12500, loss:0.15000, acc: 0.85000
[Train] Step: 13000, loss:0.15000, acc: 0.85000
[Train] Step: 13500, loss:0.12864, acc: 0.85000
[Train] Step: 14000, loss:0.10034, acc: 0.90000
[Train] Step: 14500, loss:0.15703, acc: 0.85000
[Train] Step: 15000, loss:0.23820, acc: 0.75000
(2000, 3072)
(2000,)
[Test] Step: 15000, acc: 0.81350
[Train] Step: 15500, loss:0.23760, acc: 0.75000
[Train] Step: 16000, loss:0.31062, acc: 0.70000
[Train] Step: 16500, loss:0.21815, acc: 0.75000
[Train] Step: 17000, loss:0.10000, acc: 0.90000
[Train] Step: 17500, loss:0.11240, acc: 0.85000
[Train] Step: 18000, loss:0.19982, acc: 0.80000
[Train] Step: 18500, loss:0.10140, acc: 0.90000
[Train] Step: 19000, loss:0.14327, acc: 0.80000
[Train] Step: 19500, loss:0.34986, acc: 0.65000
[Train] Step: 20000, loss:0.19997, acc: 0.80000
(2000, 3072)
(2000,)
[Test] Step: 20000, acc: 0.81400
[Train] Step: 20500, loss:0.15000, acc: 0.85000
[Train] Step: 21000, loss:0.05008, acc: 0.95000
[Train] Step: 21500, loss:0.10000, acc: 0.90000
[Train] Step: 22000, loss:0.20128, acc: 0.80000
[Train] Step: 22500, loss:0.10240, acc: 0.90000
[Train] Step: 23000, loss:0.06212, acc: 0.95000
[Train] Step: 23500, loss:0.15000, acc: 0.85000
[Train] Step: 24000, loss:0.15001, acc: 0.85000
[Train] Step: 24500, loss:0.09032, acc: 0.90000
[Train] Step: 25000, loss:0.11498, acc: 0.85000
(2000, 3072)
(2000,)
[Test] Step: 25000, acc: 0.81500
[Train] Step: 25500, loss:0.21641, acc: 0.75000
[Train] Step: 26000, loss:0.29773, acc: 0.70000
[Train] Step: 26500, loss:0.19523, acc: 0.80000
[Train] Step: 27000, loss:0.14999, acc: 0.85000
[Train] Step: 27500, loss:0.29387, acc: 0.70000
[Train] Step: 28000, loss:0.09986, acc: 0.90000
[Train] Step: 28500, loss:0.20017, acc: 0.80000
[Train] Step: 29000, loss:0.10000, acc: 0.90000
[Train] Step: 29500, loss:0.06900, acc: 0.90000
[Train] Step: 30000, loss:0.20000, acc: 0.80000
(2000, 3072)
(2000,)
[Test] Step: 30000, acc: 0.81400
[Train] Step: 30500, loss:0.29999, acc: 0.70000
[Train] Step: 31000, loss:0.10000, acc: 0.90000
[Train] Step: 31500, loss:0.28020, acc: 0.70000
[Train] Step: 32000, loss:0.23061, acc: 0.75000
[Train] Step: 32500, loss:0.09479, acc: 0.90000
[Train] Step: 33000, loss:0.15682, acc: 0.80000
[Train] Step: 33500, loss:0.20000, acc: 0.80000
[Train] Step: 34000, loss:0.10102, acc: 0.90000
[Train] Step: 34500, loss:0.10090, acc: 0.90000
[Train] Step: 35000, loss:0.20003, acc: 0.80000
(2000, 3072)
(2000,)
[Test] Step: 35000, acc: 0.81150
[Train] Step: 35500, loss:0.20290, acc: 0.80000
[Train] Step: 36000, loss:0.05014, acc: 0.95000
[Train] Step: 36500, loss:0.20018, acc: 0.80000
[Train] Step: 37000, loss:0.25043, acc: 0.75000
[Train] Step: 37500, loss:0.11776, acc: 0.85000
[Train] Step: 38000, loss:0.20055, acc: 0.80000
[Train] Step: 38500, loss:0.30000, acc: 0.70000
[Train] Step: 39000, loss:0.20100, acc: 0.80000
[Train] Step: 39500, loss:0.05023, acc: 0.95000
[Train] Step: 40000, loss:0.05000, acc: 0.95000
(2000, 3072)
(2000,)
[Test] Step: 40000, acc: 0.81500
[Train] Step: 40500, loss:0.10000, acc: 0.90000
[Train] Step: 41000, loss:0.29134, acc: 0.70000
[Train] Step: 41500, loss:0.10074, acc: 0.90000
[Train] Step: 42000, loss:0.34999, acc: 0.65000
[Train] Step: 42500, loss:0.15017, acc: 0.85000
[Train] Step: 43000, loss:0.10000, acc: 0.90000
[Train] Step: 43500, loss:0.20008, acc: 0.80000
[Train] Step: 44000, loss:0.34917, acc: 0.65000
[Train] Step: 44500, loss:0.05077, acc: 0.95000
[Train] Step: 45000, loss:0.16136, acc: 0.85000
(2000, 3072)
(2000,)
[Test] Step: 45000, acc: 0.81750
[Train] Step: 45500, loss:0.10502, acc: 0.90000
[Train] Step: 46000, loss:0.10036, acc: 0.90000
[Train] Step: 46500, loss:0.25016, acc: 0.75000
[Train] Step: 47000, loss:0.25000, acc: 0.75000
[Train] Step: 47500, loss:0.11138, acc: 0.90000
[Train] Step: 48000, loss:0.15029, acc: 0.85000
[Train] Step: 48500, loss:0.20625, acc: 0.80000
[Train] Step: 49000, loss:0.15002, acc: 0.85000
[Train] Step: 49500, loss:0.13137, acc: 0.85000
[Train] Step: 50000, loss:0.20007, acc: 0.80000
(2000, 3072)
(2000,)
[Test] Step: 50000, acc: 0.81350
[Train] Step: 50500, loss:0.25050, acc: 0.75000
[Train] Step: 51000, loss:0.16586, acc: 0.80000
[Train] Step: 51500, loss:0.05447, acc: 0.95000
[Train] Step: 52000, loss:0.09779, acc: 0.90000
[Train] Step: 52500, loss:0.09993, acc: 0.90000
[Train] Step: 53000, loss:0.10002, acc: 0.90000
[Train] Step: 53500, loss:0.24999, acc: 0.75000
[Train] Step: 54000, loss:0.09413, acc: 0.90000
[Train] Step: 54500, loss:0.05066, acc: 0.95000
[Train] Step: 55000, loss:0.06452, acc: 0.95000
(2000, 3072)
(2000,)
[Test] Step: 55000, acc: 0.81650
[Train] Step: 55500, loss:0.15000, acc: 0.85000
[Train] Step: 56000, loss:0.10002, acc: 0.90000
[Train] Step: 56500, loss:0.16470, acc: 0.80000
[Train] Step: 57000, loss:0.15289, acc: 0.85000
[Train] Step: 57500, loss:0.14925, acc: 0.85000
[Train] Step: 58000, loss:0.00004, acc: 1.00000
[Train] Step: 58500, loss:0.17873, acc: 0.80000
[Train] Step: 59000, loss:0.10021, acc: 0.90000
[Train] Step: 59500, loss:0.15005, acc: 0.85000
[Train] Step: 60000, loss:0.25000, acc: 0.75000
(2000, 3072)
(2000,)
[Test] Step: 60000, acc: 0.81350
[Train] Step: 60500, loss:0.20056, acc: 0.80000
[Train] Step: 61000, loss:0.18156, acc: 0.80000
[Train] Step: 61500, loss:0.05029, acc: 0.95000
[Train] Step: 62000, loss:0.19842, acc: 0.80000
[Train] Step: 62500, loss:0.20000, acc: 0.80000
[Train] Step: 63000, loss:0.15062, acc: 0.85000
[Train] Step: 63500, loss:0.15000, acc: 0.85000
[Train] Step: 64000, loss:0.15163, acc: 0.85000
[Train] Step: 64500, loss:0.15050, acc: 0.85000
[Train] Step: 65000, loss:0.24990, acc: 0.75000
(2000, 3072)
(2000,)
[Test] Step: 65000, acc: 0.81650
[Train] Step: 65500, loss:0.05177, acc: 0.95000
[Train] Step: 66000, loss:0.25148, acc: 0.75000
[Train] Step: 66500, loss:0.25219, acc: 0.75000
[Train] Step: 67000, loss:0.01503, acc: 1.00000
[Train] Step: 67500, loss:0.14373, acc: 0.85000
[Train] Step: 68000, loss:0.10028, acc: 0.90000
[Train] Step: 68500, loss:0.10000, acc: 0.90000
[Train] Step: 69000, loss:0.15000, acc: 0.85000
[Train] Step: 69500, loss:0.10036, acc: 0.90000
[Train] Step: 70000, loss:0.20056, acc: 0.75000
(2000, 3072)
(2000,)
[Test] Step: 70000, acc: 0.81350
[Train] Step: 70500, loss:0.10613, acc: 0.90000
[Train] Step: 71000, loss:0.20001, acc: 0.80000
[Train] Step: 71500, loss:0.00083, acc: 1.00000
[Train] Step: 72000, loss:0.05001, acc: 0.95000
[Train] Step: 72500, loss:0.25088, acc: 0.75000
[Train] Step: 73000, loss:0.05021, acc: 0.95000
[Train] Step: 73500, loss:0.05031, acc: 0.95000
[Train] Step: 74000, loss:0.23307, acc: 0.75000
[Train] Step: 74500, loss:0.20002, acc: 0.80000
[Train] Step: 75000, loss:0.15080, acc: 0.85000
(2000, 3072)
(2000,)
[Test] Step: 75000, acc: 0.81400
[Train] Step: 75500, loss:0.10167, acc: 0.90000
[Train] Step: 76000, loss:0.14995, acc: 0.85000
[Train] Step: 76500, loss:0.15019, acc: 0.85000
[Train] Step: 77000, loss:0.10022, acc: 0.90000
[Train] Step: 77500, loss:0.00061, acc: 1.00000
[Train] Step: 78000, loss:0.20023, acc: 0.80000
[Train] Step: 78500, loss:0.22279, acc: 0.75000
[Train] Step: 79000, loss:0.20009, acc: 0.80000
[Train] Step: 79500, loss:0.10502, acc: 0.90000
[Train] Step: 80000, loss:0.21905, acc: 0.75000
(2000, 3072)
(2000,)
[Test] Step: 80000, acc: 0.81250
[Train] Step: 80500, loss:0.20000, acc: 0.80000
[Train] Step: 81000, loss:0.10098, acc: 0.90000
[Train] Step: 81500, loss:0.15061, acc: 0.85000
[Train] Step: 82000, loss:0.00301, acc: 1.00000
[Train] Step: 82500, loss:0.05492, acc: 0.95000
[Train] Step: 83000, loss:0.10396, acc: 0.90000
[Train] Step: 83500, loss:0.16083, acc: 0.85000
[Train] Step: 84000, loss:0.14995, acc: 0.85000
[Train] Step: 84500, loss:0.10011, acc: 0.90000
[Train] Step: 85000, loss:0.14943, acc: 0.85000
(2000, 3072)
(2000,)
[Test] Step: 85000, acc: 0.80850
[Train] Step: 85500, loss:0.15764, acc: 0.85000
[Train] Step: 86000, loss:0.10340, acc: 0.90000
[Train] Step: 86500, loss:0.09898, acc: 0.90000
[Train] Step: 87000, loss:0.20000, acc: 0.80000
[Train] Step: 87500, loss:0.10002, acc: 0.90000
[Train] Step: 88000, loss:0.05000, acc: 0.95000
[Train] Step: 88500, loss:0.15000, acc: 0.85000
[Train] Step: 89000, loss:0.20583, acc: 0.80000
[Train] Step: 89500, loss:0.15086, acc: 0.85000
[Train] Step: 90000, loss:0.19963, acc: 0.80000
(2000, 3072)
(2000,)
[Test] Step: 90000, acc: 0.80950
[Train] Step: 90500, loss:0.20024, acc: 0.80000
[Train] Step: 91000, loss:0.20210, acc: 0.80000
[Train] Step: 91500, loss:0.10000, acc: 0.90000
[Train] Step: 92000, loss:0.20000, acc: 0.80000
[Train] Step: 92500, loss:0.10005, acc: 0.90000
[Train] Step: 93000, loss:0.34990, acc: 0.65000
[Train] Step: 93500, loss:0.12632, acc: 0.85000
[Train] Step: 94000, loss:0.20098, acc: 0.80000
[Train] Step: 94500, loss:0.15001, acc: 0.85000
[Train] Step: 95000, loss:0.15000, acc: 0.85000
(2000, 3072)
(2000,)
[Test] Step: 95000, acc: 0.81250
[Train] Step: 95500, loss:0.15014, acc: 0.85000
[Train] Step: 96000, loss:0.20001, acc: 0.80000
[Train] Step: 96500, loss:0.20008, acc: 0.80000
[Train] Step: 97000, loss:0.09981, acc: 0.90000
[Train] Step: 97500, loss:0.10011, acc: 0.90000
[Train] Step: 98000, loss:0.10008, acc: 0.90000
[Train] Step: 98500, loss:0.15029, acc: 0.85000
[Train] Step: 99000, loss:0.10099, acc: 0.90000
[Train] Step: 99500, loss:0.05000, acc: 0.95000
[Train] Step: 100000, loss:0.05000, acc: 0.95000
(2000, 3072)
(2000,)
[Test] Step: 100000, acc: 0.81150
CATALOG
  1. 1. 数据集的测试代码
  2. 2. 新增了训练的测试代码
Total : 99
2018
elasticsearch restful API hexo github 迁移 LDA 主题模型 pyspider django 搜索引擎 SAS 前端展示 数据库 前端 分支 cpu 云服务器 快速部署 bible conda pip 虚拟环境 api 接口 TF-IDF 检索模型 kibana CUDR 查询 jieba PageRank TextRank pytorch 深度学习 语言模型 excle 数据处理 xadmin 后台 快速开发 后台管理 仪式感 焦虑 军事 影评 红海行动 湄公河行动 黑鹰坠落 新时代周刊 爱情观 书评 mysql 知识库构建 NLP 词向量 Wordvec 神经网络 高血压 冠心病 尿毒症 中风 健康 关键词 EMDT 问答对 QADM pypi setuptools 问句生成 WMD RNNLM xlrd xlwt set 小猪快跑实验室 软件杯 公众号 摄影，北京，密云水库 抑郁 自我 人格 心理学 挫折 HanLP ACC wxPython 上传 下载 挂起 分词算法 HMM nginx uwsgi 静态文件 吉他，曲谱 知识库构建算法 QGDT 爬虫 self.crwal 生日 二十五 警匪 推理 三体 地球往事 番茄 鸡蛋 炒菜 挂面 早餐 IQAS 阿里云 环境部署 未知 未来 环境配置 云部署 家族 耄耋 天伦 本地连接 jpegoptim optipng 机器学习 神经元 逻辑回归 多输出 训练 梯度下降 Tensorflow 二分类 logistic 自然语言处理 自然语言处理，开发环境 NLP，预测 欧式距离，KNN 分类器，KNN NLP，KNN 机器学习，预测，sklearn 自然语言处理，NLP，实战 区块链 区块链，比特币 区块链，比特币，挖矿 区块链，比特币，共识机制 区块链，比特币，总结 restful，api restful，HTTP restful，设计要素 召回算法，协同过滤理 个性化，推荐算法 召回算法，协同过滤 usercf itemcf 分析比较，itemcf 数据集，信息抽取 算法改进，协同过滤 协同过滤 usercf，算法改进 推荐算法，在线评估 TensorFlow，在线评估 TensorFlow，mnist TensorFlow，卷积 TensorFlow，GPU 设计要素 TensorFlow mnist 卷积 比特币 总结 预测 HTTP 在线评估 GPU 召回算法 分析比较 推荐算法 个性化 KNN 挖矿 共识机制 算法改进 摄影 北京 密云水库 数据集 信息抽取 sklearn 分类器 吉他 曲谱 实战 开发环境
工作 爱好 朋友 家人