You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

5-nn-sequential-module.ipynb 178 kB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699700701702703704705706707708709710711712713714715716717718719720721722723724725726727728729730731732733734735736737738739740741742743744745746747748749750751752753754755756757758759760761762763764765766767768769770771772773774775776777778779780781782783784785786787788789790791792
  1. {
  2. "cells": [
  3. {
  4. "cell_type": "markdown",
  5. "metadata": {},
  6. "source": [
  7. "# 多层神经网络\n",
  8. "\n",
  9. "本节在前面学习线性回归和逻辑回归模型的基础上,本节学习如何利用PyTorch实现多层神经网络。"
  10. ]
  11. },
  12. {
  13. "cell_type": "markdown",
  14. "metadata": {},
  15. "source": [
  16. "## 1. 多层神经网络\n",
  17. "线性回归的公式是 $y = w x + b$, Logistic 回归的公式是 $y = Sigmoid(w x + b)$,其实它们都可以看成单层神经网络,其中 Sigmoid 被称为激活函数。"
  18. ]
  19. },
  20. {
  21. "cell_type": "markdown",
  22. "metadata": {},
  23. "source": [
  24. "### 1.1 神经网络的结构\n",
  25. "\n",
  26. "神经网络就是很多个神经元堆在一起形成一层神经网络,那么多个层堆叠在一起就是深层神经网络\n",
  27. "\n",
  28. "![nn demo](imgs/nn-forward.gif)\n",
  29. "\n",
  30. "可以看到,神经网络的结构其实非常简单,主要有输入层,隐藏层,输出层构成,输入层需要根据特征数目来决定,输出层根据解决的问题来决定,那么隐藏层的网路层数以及每层的神经元数就是可以调节的参数,而不同的层数和每层的参数对模型的影响非常大,具体的动态示例可以参考 [demo - classify2d](http://cs.stanford.edu/people/karpathy/convnetjs/demo/classify2d.html) 。神经网络向前传播也非常简单,就是一层一层不断做运算即可。"
  31. ]
  32. },
  33. {
  34. "cell_type": "markdown",
  35. "metadata": {},
  36. "source": [
  37. "### 1.2 多层神经网络示例程序\n",
  38. "\n",
  39. "首先生成一些训练、测试数据。"
  40. ]
  41. },
  42. {
  43. "cell_type": "code",
  44. "execution_count": 1,
  45. "metadata": {},
  46. "outputs": [
  47. {
  48. "data": {
  49. "image/png": "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\n",
  50. "text/plain": [
  51. "<Figure size 432x288 with 1 Axes>"
  52. ]
  53. },
  54. "metadata": {
  55. "needs_background": "light"
  56. },
  57. "output_type": "display_data"
  58. }
  59. ],
  60. "source": [
  61. "import torch\n",
  62. "import numpy as np\n",
  63. "from torch import nn\n",
  64. "from sklearn import datasets\n",
  65. "\n",
  66. "import matplotlib.pyplot as plt\n",
  67. "%matplotlib inline\n",
  68. "\n",
  69. "# generate sample data\n",
  70. "np.random.seed(0)\n",
  71. "data_x, data_y = datasets.make_moons(200, noise=0.20)\n",
  72. "\n",
  73. "# plot data\n",
  74. "plt.scatter(data_x[:, 0], data_x[:, 1], c=data_y, cmap=plt.cm.Spectral)\n",
  75. "plt.show()"
  76. ]
  77. },
  78. {
  79. "cell_type": "code",
  80. "execution_count": 2,
  81. "metadata": {},
  82. "outputs": [],
  83. "source": [
  84. "# 变量\n",
  85. "x = torch.from_numpy(data_x).float()\n",
  86. "y = torch.from_numpy(data_y).float().unsqueeze(1)\n",
  87. "\n",
  88. "\n",
  89. "# 定义两层神经网络的参数\n",
  90. "w1 = nn.Parameter(torch.randn(2, 4) * 0.1) # 隐藏层神经元个数 4\n",
  91. "b1 = nn.Parameter(torch.zeros(4))\n",
  92. "\n",
  93. "w2 = nn.Parameter(torch.randn(4, 1) * 0.1)\n",
  94. "b2 = nn.Parameter(torch.zeros(1))\n",
  95. "\n",
  96. "# 定义模型\n",
  97. "def SimpNetwork(x):\n",
  98. " x1 = torch.mm(x, w1) + b1\n",
  99. " x1 = torch.sigmoid(x1) # 使用 PyTorch 自带的 sigmoid 激活函数\n",
  100. " x2 = torch.mm(x1, w2) + b2\n",
  101. " return x2 # BCEWithLogitsLoss 已经带了sigmoid,所以此处不需要\n",
  102. "\n",
  103. "optimizer = torch.optim.SGD([w1, b1, w2, b2], 0.1)\n",
  104. "\n",
  105. "criterion = nn.BCEWithLogitsLoss()"
  106. ]
  107. },
  108. {
  109. "cell_type": "code",
  110. "execution_count": 3,
  111. "metadata": {},
  112. "outputs": [
  113. {
  114. "name": "stdout",
  115. "output_type": "stream",
  116. "text": [
  117. "epoch: 100, loss: 0.6914874315261841\n",
  118. "epoch: 200, loss: 0.6847885251045227\n",
  119. "epoch: 300, loss: 0.658918559551239\n",
  120. "epoch: 400, loss: 0.588269054889679\n",
  121. "epoch: 500, loss: 0.4917648732662201\n",
  122. "epoch: 600, loss: 0.42251646518707275\n",
  123. "epoch: 700, loss: 0.38259515166282654\n",
  124. "epoch: 800, loss: 0.3581520915031433\n",
  125. "epoch: 900, loss: 0.34184250235557556\n",
  126. "epoch: 1000, loss: 0.330547571182251\n"
  127. ]
  128. }
  129. ],
  130. "source": [
  131. "# 训练 1000 次\n",
  132. "for e in range(1000):\n",
  133. " out = SimpNetwork(x)\n",
  134. " loss = criterion(out, y)\n",
  135. " optimizer.zero_grad()\n",
  136. " loss.backward()\n",
  137. " optimizer.step()\n",
  138. " if (e + 1) % 100 == 0:\n",
  139. " print('epoch: {}, loss: {}'.format(e+1, loss.item()))"
  140. ]
  141. },
  142. {
  143. "cell_type": "code",
  144. "execution_count": 4,
  145. "metadata": {},
  146. "outputs": [],
  147. "source": [
  148. "def plot_decision_boundary(model, x, y):\n",
  149. " # Set min and max values and give it some padding\n",
  150. " x_min, x_max = x[:, 0].min() - 1, x[:, 0].max() + 1\n",
  151. " y_min, y_max = x[:, 1].min() - 1, x[:, 1].max() + 1\n",
  152. " h = 0.01\n",
  153. " # Generate a grid of points with distance h between them\n",
  154. " xx, yy = np.meshgrid(np.arange(x_min, x_max, h), np.arange(y_min, y_max, h))\n",
  155. " # Predict the function value for the whole grid .c_按行连接两个矩阵,左右相加。\n",
  156. " Z = model(np.c_[xx.ravel(), yy.ravel()])\n",
  157. " Z = Z.reshape(xx.shape)\n",
  158. " # Plot the contour and training examples\n",
  159. " plt.contourf(xx, yy, Z, cmap=plt.cm.Spectral)\n",
  160. " plt.ylabel('x2')\n",
  161. " plt.xlabel('x1')\n",
  162. " plt.scatter(x[:, 0], x[:, 1], c=y.reshape(-1), s=40, cmap=plt.cm.Spectral)"
  163. ]
  164. },
  165. {
  166. "cell_type": "code",
  167. "execution_count": 5,
  168. "metadata": {},
  169. "outputs": [
  170. {
  171. "data": {
  172. "image/png": "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\n",
  173. "text/plain": [
  174. "<Figure size 432x288 with 1 Axes>"
  175. ]
  176. },
  177. "metadata": {
  178. "needs_background": "light"
  179. },
  180. "output_type": "display_data"
  181. },
  182. {
  183. "data": {
  184. "image/png": "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\n",
  185. "text/plain": [
  186. "<Figure size 432x288 with 1 Axes>"
  187. ]
  188. },
  189. "metadata": {
  190. "needs_background": "light"
  191. },
  192. "output_type": "display_data"
  193. }
  194. ],
  195. "source": [
  196. "y_res = torch.sigmoid(SimpNetwork(x))\n",
  197. "#y_pred = np.argmax(y_res, axis=1)\n",
  198. "y_pred = (y_res > 0.5)*1\n",
  199. "\n",
  200. "# plot data\n",
  201. "plt.scatter(x[:, 0], x[:, 1], c=y, cmap=plt.cm.Spectral)\n",
  202. "plt.title(\"ground truth\")\n",
  203. "plt.show()\n",
  204. "\n",
  205. "plt.scatter(x[:, 0], x[:, 1], c=y_pred, cmap=plt.cm.Spectral)\n",
  206. "plt.title(\"predicted\")\n",
  207. "plt.show()"
  208. ]
  209. },
  210. {
  211. "cell_type": "markdown",
  212. "metadata": {},
  213. "source": [
  214. "## 2. Sequential 和 Module"
  215. ]
  216. },
  217. {
  218. "cell_type": "markdown",
  219. "metadata": {},
  220. "source": [
  221. "\n",
  222. "对于前面的线性回归模型、 Logistic回归模型和神经网络,在构建的时候定义了需要的参数。这对于比较小的模型是可行的,但是对于大的模型,比如100 层的神经网络,这个时候再去手动定义参数就显得非常麻烦,所以 PyTorch 提供了两个模块来帮助我们构建模型,一个是Sequential,一个是 Module。\n",
  223. "\n",
  224. "Sequential 允许我们构建序列化的模块,而 Module 是一种更加灵活的模型定义方式,下面分别用 `Sequential` 和 `Module` 来定义上面的神经网络。"
  225. ]
  226. },
  227. {
  228. "cell_type": "markdown",
  229. "metadata": {},
  230. "source": [
  231. "### 2.1 Sequential"
  232. ]
  233. },
  234. {
  235. "cell_type": "code",
  236. "execution_count": 6,
  237. "metadata": {},
  238. "outputs": [],
  239. "source": [
  240. "# Sequential\n",
  241. "seq_net = nn.Sequential(\n",
  242. " nn.Linear(2, 4), # PyTorch 中的线性层,wx + b\n",
  243. " nn.Tanh(),\n",
  244. " nn.Linear(4, 1)\n",
  245. ")"
  246. ]
  247. },
  248. {
  249. "cell_type": "code",
  250. "execution_count": 7,
  251. "metadata": {},
  252. "outputs": [
  253. {
  254. "data": {
  255. "text/plain": [
  256. "Linear(in_features=2, out_features=4, bias=True)"
  257. ]
  258. },
  259. "execution_count": 7,
  260. "metadata": {},
  261. "output_type": "execute_result"
  262. }
  263. ],
  264. "source": [
  265. "# 序列模块可以通过索引访问每一层\n",
  266. "seq_net[0] # 第一层"
  267. ]
  268. },
  269. {
  270. "cell_type": "code",
  271. "execution_count": 8,
  272. "metadata": {},
  273. "outputs": [
  274. {
  275. "name": "stdout",
  276. "output_type": "stream",
  277. "text": [
  278. "Parameter containing:\n",
  279. "tensor([[ 0.3485, 0.5085],\n",
  280. " [-0.6388, -0.1725],\n",
  281. " [ 0.4717, -0.2461],\n",
  282. " [-0.1726, 0.4927]], requires_grad=True)\n"
  283. ]
  284. }
  285. ],
  286. "source": [
  287. "# 打印出第一层的权重\n",
  288. "\n",
  289. "w0 = seq_net[0].weight\n",
  290. "print(w0)"
  291. ]
  292. },
  293. {
  294. "cell_type": "code",
  295. "execution_count": 9,
  296. "metadata": {},
  297. "outputs": [
  298. {
  299. "name": "stdout",
  300. "output_type": "stream",
  301. "text": [
  302. "epoch: 1000, loss: 0.3075895607471466\n",
  303. "epoch: 2000, loss: 0.3041735887527466\n",
  304. "epoch: 3000, loss: 0.30135470628738403\n",
  305. "epoch: 4000, loss: 0.25870421528816223\n",
  306. "epoch: 5000, loss: 0.14440153539180756\n",
  307. "epoch: 6000, loss: 0.10606899112462997\n",
  308. "epoch: 7000, loss: 0.09030225872993469\n",
  309. "epoch: 8000, loss: 0.08221166580915451\n",
  310. "epoch: 9000, loss: 0.0778866782784462\n",
  311. "epoch: 10000, loss: 0.07527764141559601\n"
  312. ]
  313. }
  314. ],
  315. "source": [
  316. "# generate sample data\n",
  317. "np.random.seed(0)\n",
  318. "data_x, data_y = datasets.make_moons(200, noise=0.20)\n",
  319. "\n",
  320. "# 变量\n",
  321. "x = torch.from_numpy(data_x).float()\n",
  322. "y = torch.from_numpy(data_y).float().unsqueeze(1)\n",
  323. "\n",
  324. "# 通过 parameters 可以取得模型的参数\n",
  325. "param = seq_net.parameters()\n",
  326. "\n",
  327. "# 定义优化器\n",
  328. "optim = torch.optim.SGD(param, 0.1)\n",
  329. "\n",
  330. "# 我们训练 10000 次\n",
  331. "for e in range(10000):\n",
  332. " out = seq_net(x)\n",
  333. " loss = criterion(out, y)\n",
  334. " optim.zero_grad()\n",
  335. " loss.backward()\n",
  336. " optim.step()\n",
  337. " if (e + 1) % 1000 == 0:\n",
  338. " print('epoch: {}, loss: {}'.format(e+1, loss.item()))"
  339. ]
  340. },
  341. {
  342. "cell_type": "markdown",
  343. "metadata": {},
  344. "source": [
  345. "可以看到,训练 10000 次 loss 比之前的更低,这是因为 PyTorch 自带的模块比我们写的更加稳定,同时也有一些初始化的问题在里面,关于参数初始化,我们会在后面的课程中讲到"
  346. ]
  347. },
  348. {
  349. "cell_type": "code",
  350. "execution_count": 10,
  351. "metadata": {},
  352. "outputs": [],
  353. "source": [
  354. "def plot_seq(x):\n",
  355. " out = torch.sigmoid(seq_net(torch.from_numpy(x).float())).data.numpy()\n",
  356. " out = (out > 0.5) * 1\n",
  357. " return out"
  358. ]
  359. },
  360. {
  361. "cell_type": "code",
  362. "execution_count": 11,
  363. "metadata": {},
  364. "outputs": [
  365. {
  366. "data": {
  367. "text/plain": [
  368. "Text(0.5, 1.0, 'sequential')"
  369. ]
  370. },
  371. "execution_count": 11,
  372. "metadata": {},
  373. "output_type": "execute_result"
  374. },
  375. {
  376. "data": {
  377. "image/png": "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\n",
  378. "text/plain": [
  379. "<Figure size 432x288 with 1 Axes>"
  380. ]
  381. },
  382. "metadata": {
  383. "needs_background": "light"
  384. },
  385. "output_type": "display_data"
  386. }
  387. ],
  388. "source": [
  389. "plot_decision_boundary(lambda x: plot_seq(x), x.numpy(), y.numpy())\n",
  390. "plt.title('sequential')"
  391. ]
  392. },
  393. {
  394. "cell_type": "markdown",
  395. "metadata": {},
  396. "source": [
  397. "### 2.2 保存模型参数\n",
  398. "\n",
  399. "保存模型在 PyTorch 中有两种方式,一种是将模型结构和参数都保存在一起,一种是只将参数保存下来。"
  400. ]
  401. },
  402. {
  403. "cell_type": "code",
  404. "execution_count": 19,
  405. "metadata": {
  406. "collapsed": true
  407. },
  408. "outputs": [],
  409. "source": [
  410. "# 将参数和模型保存在一起\n",
  411. "torch.save(seq_net, 'save_seq_net.pth')"
  412. ]
  413. },
  414. {
  415. "cell_type": "markdown",
  416. "metadata": {},
  417. "source": [
  418. "上面就是保存模型的方式,`torch.save`里面有两个参数,第一个是要保存的模型,第二个参数是保存的路径,读取模型的方式也非常简单"
  419. ]
  420. },
  421. {
  422. "cell_type": "code",
  423. "execution_count": 20,
  424. "metadata": {
  425. "collapsed": true
  426. },
  427. "outputs": [],
  428. "source": [
  429. "# 读取保存的模型\n",
  430. "seq_net1 = torch.load('save_seq_net.pth')"
  431. ]
  432. },
  433. {
  434. "cell_type": "code",
  435. "execution_count": 21,
  436. "metadata": {},
  437. "outputs": [
  438. {
  439. "data": {
  440. "text/plain": [
  441. "Sequential(\n",
  442. " (0): Linear(in_features=2, out_features=4, bias=True)\n",
  443. " (1): Tanh()\n",
  444. " (2): Linear(in_features=4, out_features=1, bias=True)\n",
  445. ")"
  446. ]
  447. },
  448. "execution_count": 21,
  449. "metadata": {},
  450. "output_type": "execute_result"
  451. }
  452. ],
  453. "source": [
  454. "seq_net1"
  455. ]
  456. },
  457. {
  458. "cell_type": "code",
  459. "execution_count": 22,
  460. "metadata": {},
  461. "outputs": [
  462. {
  463. "name": "stdout",
  464. "output_type": "stream",
  465. "text": [
  466. "Parameter containing:\n",
  467. "tensor([[-5.7823, 5.7006],\n",
  468. " [ 5.3129, 3.6949],\n",
  469. " [ 3.5471, -0.7431],\n",
  470. " [ 2.4003, 1.7605]], requires_grad=True)\n"
  471. ]
  472. }
  473. ],
  474. "source": [
  475. "print(seq_net1[0].weight)"
  476. ]
  477. },
  478. {
  479. "cell_type": "markdown",
  480. "metadata": {},
  481. "source": [
  482. "我们可以看到我们重新读入了模型,并且将其命名为 seq_net1,并且打印了第一层的参数\n",
  483. "\n",
  484. "下面我们看看第二种保存模型的方式,只保存参数而不保存模型结构"
  485. ]
  486. },
  487. {
  488. "cell_type": "code",
  489. "execution_count": 23,
  490. "metadata": {
  491. "collapsed": true
  492. },
  493. "outputs": [],
  494. "source": [
  495. "# 保存模型参数\n",
  496. "torch.save(seq_net.state_dict(), 'save_seq_net_params.pth')"
  497. ]
  498. },
  499. {
  500. "cell_type": "markdown",
  501. "metadata": {},
  502. "source": [
  503. "通过上面的方式,我们保存了模型的参数,如果要重新读入模型的参数,首先我们需要重新定义一次模型,接着重新读入参数"
  504. ]
  505. },
  506. {
  507. "cell_type": "code",
  508. "execution_count": 24,
  509. "metadata": {},
  510. "outputs": [
  511. {
  512. "data": {
  513. "text/plain": [
  514. "<All keys matched successfully>"
  515. ]
  516. },
  517. "execution_count": 24,
  518. "metadata": {},
  519. "output_type": "execute_result"
  520. }
  521. ],
  522. "source": [
  523. "seq_net2 = nn.Sequential(\n",
  524. " nn.Linear(2, 4),\n",
  525. " nn.Tanh(),\n",
  526. " nn.Linear(4, 1)\n",
  527. ")\n",
  528. "\n",
  529. "seq_net2.load_state_dict(torch.load('save_seq_net_params.pth'))"
  530. ]
  531. },
  532. {
  533. "cell_type": "code",
  534. "execution_count": 25,
  535. "metadata": {},
  536. "outputs": [
  537. {
  538. "data": {
  539. "text/plain": [
  540. "Sequential(\n",
  541. " (0): Linear(in_features=2, out_features=4, bias=True)\n",
  542. " (1): Tanh()\n",
  543. " (2): Linear(in_features=4, out_features=1, bias=True)\n",
  544. ")"
  545. ]
  546. },
  547. "execution_count": 25,
  548. "metadata": {},
  549. "output_type": "execute_result"
  550. }
  551. ],
  552. "source": [
  553. "seq_net2"
  554. ]
  555. },
  556. {
  557. "cell_type": "code",
  558. "execution_count": 26,
  559. "metadata": {},
  560. "outputs": [
  561. {
  562. "name": "stdout",
  563. "output_type": "stream",
  564. "text": [
  565. "Parameter containing:\n",
  566. "tensor([[-5.7823, 5.7006],\n",
  567. " [ 5.3129, 3.6949],\n",
  568. " [ 3.5471, -0.7431],\n",
  569. " [ 2.4003, 1.7605]], requires_grad=True)\n"
  570. ]
  571. }
  572. ],
  573. "source": [
  574. "print(seq_net2[0].weight)"
  575. ]
  576. },
  577. {
  578. "cell_type": "markdown",
  579. "metadata": {},
  580. "source": [
  581. "通过这种方式我们也重新读入了相同的模型,打印第一层的参数对比,发现和前面的办法是一样"
  582. ]
  583. },
  584. {
  585. "cell_type": "markdown",
  586. "metadata": {},
  587. "source": [
  588. "有这两种保存和读取模型的方法,我们推荐使用**第二种**,因为第二种可移植性更强"
  589. ]
  590. },
  591. {
  592. "cell_type": "markdown",
  593. "metadata": {},
  594. "source": [
  595. "### 2.3 Module\n",
  596. "\n",
  597. "下面再用 Module 定义这个模型,下面是使用 Module 的模板\n",
  598. "\n",
  599. "```\n",
  600. "class 网络名字(nn.Module):\n",
  601. " def __init__(self, 一些定义的参数):\n",
  602. " super(网络名字, self).__init__()\n",
  603. " self.layer1 = nn.Linear(num_input, num_hidden)\n",
  604. " self.layer2 = nn.Sequential(...)\n",
  605. " ...\n",
  606. " \n",
  607. " 定义需要用的网络层\n",
  608. " \n",
  609. " def forward(self, x): # 定义前向传播\n",
  610. " x1 = self.layer1(x)\n",
  611. " x2 = self.layer2(x)\n",
  612. " x = x1 + x2\n",
  613. " ...\n",
  614. " return x\n",
  615. "```\n",
  616. "\n",
  617. "注意的是,Module 里面也可以使用 Sequential,同时 Module 非常灵活,具体体现在 forward 中,如何复杂的操作都能直观的在 forward 里面执行\n",
  618. "\n",
  619. "下面我们照着模板实现一下上面的神经网络"
  620. ]
  621. },
  622. {
  623. "cell_type": "code",
  624. "execution_count": 12,
  625. "metadata": {},
  626. "outputs": [],
  627. "source": [
  628. "class SimpNet(nn.Module):\n",
  629. " def __init__(self, num_input, num_hidden, num_output):\n",
  630. " super(SimpNet, self).__init__()\n",
  631. " self.layer1 = nn.Linear(num_input, num_hidden)\n",
  632. " \n",
  633. " self.layer2 = nn.Tanh()\n",
  634. " \n",
  635. " self.layer3 = nn.Linear(num_hidden, num_output)\n",
  636. " \n",
  637. " def forward(self, x):\n",
  638. " x = self.layer1(x)\n",
  639. " x = self.layer2(x)\n",
  640. " x = self.layer3(x)\n",
  641. " return x"
  642. ]
  643. },
  644. {
  645. "cell_type": "code",
  646. "execution_count": 13,
  647. "metadata": {},
  648. "outputs": [],
  649. "source": [
  650. "mo_net = SimpNet(2, 4, 1)"
  651. ]
  652. },
  653. {
  654. "cell_type": "code",
  655. "execution_count": 14,
  656. "metadata": {},
  657. "outputs": [
  658. {
  659. "name": "stdout",
  660. "output_type": "stream",
  661. "text": [
  662. "Linear(in_features=2, out_features=4, bias=True)\n"
  663. ]
  664. }
  665. ],
  666. "source": [
  667. "# 访问模型中的某层可以直接通过名字\n",
  668. "\n",
  669. "# 第一层\n",
  670. "l1 = mo_net.layer1\n",
  671. "print(l1)"
  672. ]
  673. },
  674. {
  675. "cell_type": "code",
  676. "execution_count": 15,
  677. "metadata": {},
  678. "outputs": [
  679. {
  680. "name": "stdout",
  681. "output_type": "stream",
  682. "text": [
  683. "Parameter containing:\n",
  684. "tensor([[ 0.6988, 0.2605],\n",
  685. " [-0.4452, 0.1708],\n",
  686. " [-0.3578, 0.6637],\n",
  687. " [ 0.2984, -0.1281]], requires_grad=True)\n"
  688. ]
  689. }
  690. ],
  691. "source": [
  692. "# 打印出第一层的权重\n",
  693. "print(l1.weight)"
  694. ]
  695. },
  696. {
  697. "cell_type": "code",
  698. "execution_count": 16,
  699. "metadata": {},
  700. "outputs": [],
  701. "source": [
  702. "# 定义优化器\n",
  703. "optim = torch.optim.SGD(mo_net.parameters(), 1.)"
  704. ]
  705. },
  706. {
  707. "cell_type": "code",
  708. "execution_count": 17,
  709. "metadata": {},
  710. "outputs": [
  711. {
  712. "name": "stdout",
  713. "output_type": "stream",
  714. "text": [
  715. "epoch: 1000, loss: 0.0754304826259613\n",
  716. "epoch: 2000, loss: 0.06512685120105743\n",
  717. "epoch: 3000, loss: 0.061497319489717484\n",
  718. "epoch: 4000, loss: 0.055132776498794556\n",
  719. "epoch: 5000, loss: 0.04916892945766449\n",
  720. "epoch: 6000, loss: 0.04603230580687523\n",
  721. "epoch: 7000, loss: 0.04394793137907982\n",
  722. "epoch: 8000, loss: 0.04242979362607002\n",
  723. "epoch: 9000, loss: 0.041267599910497665\n",
  724. "epoch: 10000, loss: 0.04034609720110893\n"
  725. ]
  726. }
  727. ],
  728. "source": [
  729. "# 我们训练 10000 次\n",
  730. "for e in range(10000):\n",
  731. " out = mo_net(x)\n",
  732. " loss = criterion(out, y)\n",
  733. " optim.zero_grad()\n",
  734. " loss.backward()\n",
  735. " optim.step()\n",
  736. " if (e + 1) % 1000 == 0:\n",
  737. " print('epoch: {}, loss: {}'.format(e+1, loss.item()))"
  738. ]
  739. },
  740. {
  741. "cell_type": "code",
  742. "execution_count": 33,
  743. "metadata": {
  744. "collapsed": true
  745. },
  746. "outputs": [],
  747. "source": [
  748. "# 保存模型\n",
  749. "torch.save(mo_net.state_dict(), 'module_net.pth')"
  750. ]
  751. },
  752. {
  753. "cell_type": "markdown",
  754. "metadata": {},
  755. "source": [
  756. "可以看到我们得到了相同的结果,而且使用 Sequential 和 Module 来定义模型更加方便\n",
  757. "\n",
  758. "在这一节中我们还是使用梯度下降法来优化参数,在神经网络中,这种优化方法有一个特别的名字,反向传播算法,下一次课我们会讲一讲什么是反向传播算法"
  759. ]
  760. },
  761. {
  762. "cell_type": "markdown",
  763. "metadata": {},
  764. "source": [
  765. "## 练习题\n",
  766. "\n",
  767. "* 改变网络的隐藏层神经元数目,或者试试定义一个 5 层甚至更深的模型,增加训练次数,改变学习率,看看结果会怎么样"
  768. ]
  769. }
  770. ],
  771. "metadata": {
  772. "kernelspec": {
  773. "display_name": "Python 3 (ipykernel)",
  774. "language": "python",
  775. "name": "python3"
  776. },
  777. "language_info": {
  778. "codemirror_mode": {
  779. "name": "ipython",
  780. "version": 3
  781. },
  782. "file_extension": ".py",
  783. "mimetype": "text/x-python",
  784. "name": "python",
  785. "nbconvert_exporter": "python",
  786. "pygments_lexer": "ipython3",
  787. "version": "3.9.7"
  788. }
  789. },
  790. "nbformat": 4,
  791. "nbformat_minor": 2
  792. }

机器学习越来越多应用到飞行器、机器人等领域,其目的是利用计算机实现类似人类的智能,从而实现装备的智能化与无人化。本课程旨在引导学生掌握机器学习的基本知识、典型方法与技术,通过具体的应用案例激发学生对该学科的兴趣,鼓励学生能够从人工智能的角度来分析、解决飞行器、机器人所面临的问题和挑战。本课程主要内容包括Python编程基础,机器学习模型,无监督学习、监督学习、深度学习基础知识与实现,并学习如何利用机器学习解决实际问题,从而全面提升自我的《综合能力》。