This commit is contained in:
ha4ska committed 2026-10-29 23:24:10 +05:00
1 parent d049732103
commit 22a5be5dc9
1 file changed
+244
+244
View File
@@ -0,0 +1,244 @@
{
"cells": [
{
"cell_type": "code",
"execution_count": 5,
"id": "cc8ac5b2",
"metadata": {},
"outputs": [
{
"name": "stdout",
"output_type": "stream",
"text": [
"--- AND ---\n",
"Веса: [0.11624345 0.09388244]\n",
"Смещение: -0.2\n",
"Предсказание: [0 0 0 1]\n",
"Ошибки по эпохам: [2, 3, 1, 0]\n",
"\n",
"--- OR ---\n",
"Веса: [0.11624345 0.19388244]\n",
"Смещение: -0.1\n",
"Предсказание: [0 1 1 1]\n",
"Ошибки по эпохам: [2, 2, 2, 1, 0]\n",
"\n",
"--- XOR ---\n",
"Веса: [-0.08375655 -0.00611756]\n",
"Смещение: 0.0\n",
"Предсказание: [1 0 0 0]\n",
"Ошибки по эпохам: [3, 3, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4, 4]\n",
"\n"
]
}
],
"source": [
"import numpy as np\n",
"import matplotlib.pyplot as plt\n",
"\n",
"class Perceptron:\n",
" def __init__(self, eta=0.1, n_iter=100, random_state=1):\n",
" self.eta = eta\n",
" self.n_iter = n_iter\n",
" self.random_state = random_state\n",
"\n",
" def fit(self, X, y):\n",
" rgen = np.random.RandomState(self.random_state)\n",
" self.w_ = rgen.normal(loc=0.0, scale=0.01, size=X.shape[1])\n",
" self.b_ = 0.0\n",
" self.errors_ = []\n",
"\n",
" for _ in range(self.n_iter):\n",
" errors = 0\n",
" for xi, target in zip(X, y):\n",
" y_pred = 1 if self.net_input(xi) >= 0.0 else 0\n",
" update = self.eta * (target - y_pred)\n",
"\n",
" self.w_ += update * xi\n",
" self.b_ += update\n",
"\n",
" if update != 0.0:\n",
" errors += 1\n",
"\n",
" self.errors_.append(errors)\n",
" if errors == 0:\n",
" break\n",
"\n",
" return self\n",
"\n",
" def net_input(self, X):\n",
" return np.dot(X, self.w_) + self.b_\n",
"\n",
" def predict(self, X):\n",
" return np.where(self.net_input(X) >= 0.0, 1, 0)\n",
"\n",
"\n",
"X = np.array([\n",
" [0,0],\n",
" [0,1],\n",
" [1,0],\n",
" [1,1]\n",
"])\n",
"\n",
"y_and = np.array([0,0,0,1])\n",
"y_or = np.array([0,1,1,1])\n",
"y_xor = np.array([0,1,1,0])\n",
"\n",
"for name, y in [('AND', y_and), ('OR', y_or), ('XOR', y_xor)]:\n",
" ppn = Perceptron(eta=0.1, n_iter=20, random_state=1)\n",
" ppn.fit(X,y)\n",
"\n",
" print(f'--- {name} ---')\n",
" print('Веса:', ppn.w_)\n",
" print('Смещение:', ppn.b_)\n",
" print('Предсказание:', ppn.predict(X))\n",
" print('Ошибки по эпохам:', ppn.errors_)\n",
" print()"
]
},
{
"cell_type": "code",
"execution_count": 7,
"id": "f4f788a1",
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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 truncated
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"ppn = Perceptron(eta=0.1, n_iter=20, random_state=1)\n",
"ppn.fit(X, y_xor)\n",
"\n",
"plt.plot(range(1, len(ppn.errors_) + 1), ppn.errors_, marker='o')\n",
"plt.xlabel(\"Epoch\")\n",
"plt.ylabel(\"Error figures\")\n",
"plt.title(\"Обучение перцептрона для OR\")\n",
"plt.grid(True)\n",
"plt.show()"
]
},
{
"cell_type": "code",
"execution_count": 10,
"id": "02a790aa",
"metadata": {},
"outputs": [
{
"data": {
"image/png": "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 truncated
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"def plot_decision_boundary(X,y,model):\n",
"\n",
" x_min, x_max = X[:, 0].min() - 0.5, X[:, 0].max() + 0.5\n",
" y_min, y_max = X[:, 1].min() - 0.5, X[:, 1].max() + 0.5\n",
"\n",
" xx, yy = np.meshgrid(\n",
" np.arange(x_min, x_max, 0.02),\n",
" np.arange(y_min, y_max, 0.02)\n",
" )\n",
"\n",
" Z = model.predict(np.c_[xx.ravel(), yy.ravel()])\n",
" Z = Z.reshape(xx.shape)\n",
"\n",
" plt.contourf(xx, yy, Z, alpha=0.3, cmap='bwr')\n",
" plt.scatter(X[:,0], X[:,1], c=y, cmap='bwr', edgecolors='k')\n",
" plt.xlabel('x1')\n",
" plt.ylabel('x2')\n",
" plt.title(\"Разделяющая граница\")\n",
"\n",
" plt.show()\n",
"\n",
"ppn = Perceptron(eta=0.1, n_iter=20, random_state=1)\n",
"ppn.fit(X, y_or)\n",
"plot_decision_boundary(X, y_or, ppn)"
]
},
{
"cell_type": "code",
"execution_count": 11,
"id": "30f0d84c",
"metadata": {},
"outputs": [
{
"data": {
"image/png": "iVBORw0KGgoAAAANSUhEUgAAAjYAAAHFCAYAAADhWLMfAAAAOnRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjExLjIsIGh0dHBzOi8vbWF0cGxvdGxpYi5vcmcvgI3uAAAAAAlwSFlzAAAPYQAAD2EBqD+naQAAjO1JREFUeJzt3Qd4U2UbBuAnJy1ll733HrJkT9mKjAqyERFQcbBkyBIQlOWPiAICskHZCEX23rL3XrL3aqEU2uac/3q/JiWUtLSlbdr0ua8rIslpcpqU5s33vcNkGIYBIiIiIhegOfsEiIiIiKILAxsiIiJyGQxsiIiIyGUwsCEiIiKXwcCGiIiIXAYDGyIiInIZDGyIiIjIZTCwISIiIpfBwIaIiIhcBgMbojfUpk0buLm5hVySJUuGQoUKoV+/fnj8+DGfXyKiWOQWmw9G5IosFou6PHv2DGazGT4+PliyZAm++uor/Pvvv9i8eTNMJpOzT5OIKEHgig1RNJGgRlZs0qZNi88//xxt27bF1q1bceDAAT7HRESxhIENUQwpUqSI+vPKlSvqz44dO4ZsVyVKlAiZMmVCkyZNcPz48Ze+7vDhw2jVqhWyZMmCdOnSoUqVKli6dKnDx0iePPlL22C2y/Lly185dsuWLXjvvfdU4JUiRQpUrFgRK1aseOW4M2fOOLxPuVy+fDnkOFml+vHHH9X3Kfdnf9yYMWNCjqtfvz4qVKgQ8nd/f38MHz4cJUqUgKenJzJkyIBmzZrh1KlTL52HfN+1atV66bq///5b3f+ePXteul62/j788MOXrovo8x3R78MROUdHz5NcH/r7OH/+vHr+U6ZMiRw5cmDgwIEICgqK0jk7em7EqlWr1NdKQG1/7LvvvvvScRcvXgz5Xk+cOBGl+43M+RLFJgY2RDFEAgSRLVs29eeUKVPUdpVcZLtq3bp1CAgIQJ06dXD//v2Qr/vhhx/QunVrHD16VL3ZN27cWL1p7969+5XHkDfGr7/+OuR+5Y1H3qh1XX/puLlz56o3rAIFCmDfvn3qjU3egD744AMsXLjwpWMNw1D38ccff4Tc72+//aauk9tsRo0ahcGDB+Pbb7/F9evX1XHyZ+jHl7/b3sDlevl+5Os6dOiAc+fOYePGjbh9+7YKfuyDG/ka+Vp78vWhzyOsYyP6fEf0+3BEHrdy5cohj2P/Otgf4+vri86dO6vH+e+//1Qg9fPPP6uVvaics6PvN6znx9Gx8jMjAWZEjg3veY/o+RLFJgY2RNHMz88Pf/75J2bPno3y5cujTJkywf/YNC3k022SJElQvHhxzJkzB7du3YK3t3fI10t+TsOGDdVqjSQiy6qMvKE4+hQcGBgId3f3kPuV7TBH59OlSxe8//77KkDJkycP0qdPj969e6NFixbqT/s3LLlP4eHhEXK/cu6hbd++Hbly5cInn3yiViHCenx7mzZtwtq1a/HFF1+gW7duarWmWLFiWLZsmXrzHDBgAKJLRJ/vqHwf9iR/KvSKTeiv379/vwqgZJVMVsw+/vhjlVw+Y8aMl1ZMInrOb2LBggXqNZDg5k3FxvkSRRYDG6JokjhxYvULXrZX5A26U6dOWL16dUhQIL/sJaE4b968SJo0qTpWAgwh2xT2ZLXEVmElQUC5cuXQtGnTl46RT9wSDMgbSnh27NiBBw8eqCAmNPlkLVtlsopgf7/idfcrAZJsTc2bNw/Pnz8P91jJM5Lvx7YlEnrbKE2aNGpFacOGDYguEX2+I/N9RFXu3LnV1ps9WTET9t9zZH5GZFUodEDl5eUV7nnIytE333yjVookyHpTkTlfotjCwIYomsjKiCzJy4qHvFGOGzcOqVOnVrfJddWrV1crFlOnTlVbHRJAPHnyRH3il+V7e5999pm6Lwk6JEiS7RtZubEn9yEkr+F1bz5Ctn4k+JKVGMmHkMunn36qbrt3717I8bYtBFlNCc/QoUPRoEEDlSRtC+rCOpfSpUur72fixInq75kzZ37lGPlaKY+PjuAiMs93ZL6PqMqYMeMr19kew/bcR/ZnpFq1aq9sgYWVi2XTv39/FQyPHDnyjb+nyJ4vUWxhuTdRNFdFOSL5MZJzs3jxYtSoUSPkeglcQuct2G9vZM+eXeXcSJ6OHCfbFzaSg2NLnA2PbH3YtiBki8sR+/OW/BuRP3/+196v3GfZsmWRL18+lasjQVFYQYE8RtGiRdX/S05NwYIFX7pdrpMVKgm83lRknu/Ifh9RcefOnVeuk+/X9viRPWf7nxF7jrYMbSS3SgJL2SpKlSrVG30/UTlfotjCFRuiWGD7RR96e0feZCLytZK4eejQoZeuX7Nmjbq/120pyCd7yR2xVRQ5uoTe4ihcuLDDVYbQ+vbtq1aEJk+eHKHcFAkepCIodP6FJJ5Kv5+aNWvCGc93ZL+PyJKtvpMnT750na1yzRYUvMnPyOvIz49sjcrzK4np0SEmz5foTTCwIYoFb7/9ttp+kWqYS5cuqVwHWb6XxFH7N1HJRZGkWqkOkqV+OVa2kOTTvWyX2Gzbtk0lJ3/00UcqtyE8EtT8+uuvqjKqR48eakXGdt/yJiTbXEK2EOQ4SV7+7rvvXvs9SQKqHC/BgC2v4nVk+0tKvSdMmKAqamTrSXIxmjdvrs5Jnp/YfL6j+n1EVsmSJVWStpTyP336VK1yyPPQsmXLkNybyJxzZEmCtPxM2bYCo0NMni/Rm2BgQxQLJD9m5cqVajVA+qVIFc7evXtVVYw9eZOTKiH5VC1fI7kpFy5cUEGJVNIIqWx65513VC7KtGnTHPZPkeRc+wojqfiRsurTp0+rNyQJhiRZV/Ijvv/+e3VM+/bt1TaMBDav+1R/9+5ddZ+Sl2ILjCIzgmL69OkqB0mShkuVKqXeCCXJWapqwkuQtSVAh+4fI8+RrIDYSusj+ny/yfcRGRJc/vTTT6o6TfKuvvzyS5VHNXPmzJBjInrOUV2xkfwa2Wp7nfCSkuVnRr6PmD5fojdhMrgZSvRGJBlTLmHl10TkTUfyJcLLj7A3duxYzJ8/XwUCYZHARraToiNJ1H47zPY92v4uAUnocRHSC8X+enlu5PjIfop31DclPBLYSIWOLVk6Is93ZL6PsO5LhPe9SX8eSUqWBolRFfpnJKzHdfT92J7H0D+foV9T+2PDIoFNvXr11NZdZM6XKDYxeZjoDckv7zf5BR7VZfvwAqnonk0VOlHVUeJqWOcV1ecmss9LRB/H/n4j83287r5iUujHCetxHX0/0XGs/ddE5XyJYhNXbIjiGflELasg4b152DrmJqRPzBF5XpwhOlZs4gpZiZGfKQ51pbiMKzZE8Yy8qbzuzTshBTSReV7ozfD5pfiAKzZERDEoInk4RBR9GNgQERGRy0h469VERETkshJUjo0kFt64cQMpUqRg8hsREVE8Kg6Qhp5ZsmR5bQ5hggpsJKiR2TtEREQU/1y9ejWkEWdYElRgIys14uq8eUj5mjb0REREFDf4Pn2K7K1ahbyPhydBBTa23gsS1KRMlszZp0NERESREJEeSkweJiIiIpfBwIaIiIhcBgMbIiIichkMbIiIiMhlMLAhIiIil8HAhoiIiFwGAxsiIiJyGQxsiIiIyGUwsCEiIiKXwcCGiIiIXAYDGyIiInIZDGyIiIjIZTCwISIiIpfBwIaIiIhcBgMbIiIichkMbIiIiMhlMLAhIiIil8HAhoiIiFwGAxsiIiJyGQxsiIiIyGUwsCEiIiKXwcCGiIiIXAYDGyIiInIZDGyIiIjIZTCwISIiIpfBwIaIiIhcBgMbIiIichluzj4BIiKiyHgeEIClO3fiyIULSJwoERpWrIi38+fnk0gKAxsiIoo3Nh48iNY//og7vr7I7eYGX8PA97Nno26pUpg/aBBSp0jh7FMkJ+NWFBERxQtHL15Eg/79UfLxY5wCcDEoCLcsFiwGcODIETQaMACGYTj7NMnJGNgQEVG8MGrePGQ1DHgbBgrZbTt8CGCermPHyZPYdOiQk8+SnI2BDRERxXm6rmPxtm34zGJBYge31wZQwGzGgi1bnHB2FJcwsCEiojjveWAgAiwWZA3jdhOAbIYB36dPY/nMKK5hYENERHGeVD9lS5MG28O43Q/Aflm1yZYtls+M4pp4FdgcP34cX375JapUqYIGDRpg4sSJCAwMdPZpERFRDDOZTPi8USPM0TQcDnWbpAv/COCJYaBjvXp8LRK4eFPuferUKXz22Wf49NNP8fHHH+P8+fPo06cPDh48iClTpjj79IiIKIZ98+GHWL59O6r99x++0nVICPMAwDSTCSsNA6M+/RQ5M2Z86WssFgtW7N6NmWvX4ubdu8iULh3avfuu6n3jZjbzNXNBJiOe1MYFBATA3d1dRe02o0ePxogRI3D//v0I3Yevry88PT3h4+2NlMmSxeDZEhFRTPD181N9a6avWgUff391XfGcOdHvo4/QskaNl459+uwZPhg4EOsPHUJZTUMxXccJTcMeXUeN4sWxfNgwJE+ShC9UPHndPb284OPjg5QpU7pGYBPa8+fP0bRpUxXwrF27NkJfw8CGiMg1+D9/jit37qjcmxwZMrz0odfm859/xl9r12KZrqOO3fWbAHhpGprWqoUZffrE6nlTzAc28SrHRnzxxRcoXrw4MmTIoPJrFi1aFG7wI8GM/YWIiOK/JB4eKJg9u9p6chTU3PPxwax16zA4VFAjagIYquv4a9Mm3Hogm1nkSuJdYNO7d29MmzYNY8eOxbFjxzBgwIAwj5VtKtl6sl2yZ88eq+dKRETOseP4cVUe3iaM2+X6QIsF244ejeUzo5gW7wKbvHnzomzZsmjfvj0mTJiA8ePH49q1aw6P7devn1q2sl2uXr0a6+dLRESxT5KGRaIwbrddb9H1WDsnih3xLrCxlz59evXno0ePHN7u4eGh9uLsL0RE5PrKFSoEzWTC0jBu/9va1K9C4cKxfGYU0+JNYLN48WLLine truncated
"text/plain": [
"<Figure size 640x480 with 1 Axes>"
]
},
"metadata": {},
"output_type": "display_data"
}
],
"source": [
"from sklearn.datasets import make_blobs\n",
"\n",
"X_blobs, y_blobs = make_blobs(\n",
" n_samples=100,\n",
" centers=[[0,0], [2,2]],\n",
" cluster_std=0.5,\n",
" random_state=1\n",
")\n",
"\n",
"ppn = Perceptron(eta=0.1, n_iter=50, random_state=1)\n",
"ppn.fit(X_blobs, y_blobs)\n",
"\n",
"plot_decision_boundary(X_blobs, y_blobs, ppn)"
]
},
{
"cell_type": "markdown",
"id": "0722466d",
"metadata": {},
"source": [
"## Контрольные вопросы\n",
"\n",
"1. Однослойный перцептрон - самая простая модель искуственной нейронной сети, предназначенная для решения задачи бинарной классификации (разделение на два класса)\n",
"2. Сначала ищет взвешенную сумму весов (w) и входноых данных (x) и передает их в функцию активации.\n",
"3. Используется жёсткая пороговая функция. f(z) = 1 если z>=0, 0 если z<0\n",
"4. Как и всегда - вычисляется предсказание, вычислается ошибка, веса и смещения корректируются.\n",
"5. Скорость обучения это параметр, который задает величину шага корректировки весов на каждой ошибке. Черезмерно большой шаг ведет к расходимости, слишком маленький к увеличению времени обучения и риску застрять в области неоптимальных значений\n",
"6. Потому что однослойный перцептрон может построить только одну разделяющую границу (как на графике сверху). В XOR невохможно разделить 0,1 1,0 от 1,1 0,0 одной линией.\n",
"7. Если данные не линейно разделимы то перцептрон не сойдется, т.е цикл обновления весов бут бесконечным.\n",
"8. Смещение позволяет двигаться относительно начала координат ( в нашем случае сдвиг вверх), если не будет смещения то прямая будет проходить точно через центр координат (0,0), а если классы будут требовать разделения не от начала координат, то перцептрон не сможет их разделить.\n",
"9. если на обучении не будет ошибок, провести вычисление метрик Acc, Precision и Recall с F1 мерой для валидационных (тестовых) данных, которых модель не видела.\n"
]
}
],
"metadata": {
"kernelspec": {
"display_name": "ai_lesson",
"language": "python",
"name": "python3"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.12.15"
}
},
"nbformat": 4,
"nbformat_minor": 5
}