Spaces:
Sleeping
Sleeping
Tweak PySR demo
Browse files- examples/pysr_demo.ipynb +11 -7
examples/pysr_demo.ipynb
CHANGED
|
@@ -990,6 +990,7 @@
|
|
| 990 |
]
|
| 991 |
},
|
| 992 |
{
|
|
|
|
| 993 |
"cell_type": "markdown",
|
| 994 |
"metadata": {
|
| 995 |
"id": "3hS2kTAbbDhL"
|
|
@@ -999,9 +1000,9 @@
|
|
| 999 |
"\n",
|
| 1000 |
"Let's consider a time series problem:\n",
|
| 1001 |
"\n",
|
| 1002 |
-
"$$ z = y^2,\\quad y = \\frac{1}{
|
| 1003 |
"\n",
|
| 1004 |
-
"Imagine our time series is
|
| 1005 |
"\n",
|
| 1006 |
"But, as in our paper, **we can break this problem down into parts with a neural network. Then approximate the neural network with the symbolic regression!**\n",
|
| 1007 |
"\n",
|
|
@@ -1018,7 +1019,7 @@
|
|
| 1018 |
"source": [
|
| 1019 |
"###### np.random.seed(0)\n",
|
| 1020 |
"N = 100000\n",
|
| 1021 |
-
"Nt =
|
| 1022 |
"X = 6 * np.random.rand(N, Nt, 5) - 3\n",
|
| 1023 |
"y_i = X[..., 0] ** 2 + 6 * np.cos(2 * X[..., 2])\n",
|
| 1024 |
"y = np.sum(y_i, axis=1) / y_i.shape[1]\n",
|
|
@@ -1299,6 +1300,7 @@
|
|
| 1299 |
]
|
| 1300 |
},
|
| 1301 |
{
|
|
|
|
| 1302 |
"cell_type": "markdown",
|
| 1303 |
"metadata": {
|
| 1304 |
"id": "6WuaeqyqbDhe"
|
|
@@ -1306,7 +1308,7 @@
|
|
| 1306 |
"source": [
|
| 1307 |
"Recall we are searching for $y_i$ above:\n",
|
| 1308 |
"\n",
|
| 1309 |
-
"$$ z = y^2,\\quad y = \\frac{1}{
|
| 1310 |
]
|
| 1311 |
},
|
| 1312 |
{
|
|
@@ -1373,11 +1375,13 @@
|
|
| 1373 |
},
|
| 1374 |
"gpuClass": "standard",
|
| 1375 |
"kernelspec": {
|
| 1376 |
-
"display_name": "Python
|
| 1377 |
-
"
|
|
|
|
| 1378 |
},
|
| 1379 |
"language_info": {
|
| 1380 |
-
"name": "python"
|
|
|
|
| 1381 |
}
|
| 1382 |
},
|
| 1383 |
"nbformat": 4,
|
|
|
|
| 990 |
]
|
| 991 |
},
|
| 992 |
{
|
| 993 |
+
"attachments": {},
|
| 994 |
"cell_type": "markdown",
|
| 995 |
"metadata": {
|
| 996 |
"id": "3hS2kTAbbDhL"
|
|
|
|
| 1000 |
"\n",
|
| 1001 |
"Let's consider a time series problem:\n",
|
| 1002 |
"\n",
|
| 1003 |
+
"$$ z = y^2,\\quad y = \\frac{1}{10} \\sum(y_i),\\quad y_i = x_{i0}^2 + 6 \\cos(2*x_{i2})$$\n",
|
| 1004 |
"\n",
|
| 1005 |
+
"Imagine our time series is 10 timesteps. That is very hard for symbolic regression, even if we impose the inductive bias of $$z=f(\\sum g(x_i))$$ - it is the square of the number of possible equations!\n",
|
| 1006 |
"\n",
|
| 1007 |
"But, as in our paper, **we can break this problem down into parts with a neural network. Then approximate the neural network with the symbolic regression!**\n",
|
| 1008 |
"\n",
|
|
|
|
| 1019 |
"source": [
|
| 1020 |
"###### np.random.seed(0)\n",
|
| 1021 |
"N = 100000\n",
|
| 1022 |
+
"Nt = 10\n",
|
| 1023 |
"X = 6 * np.random.rand(N, Nt, 5) - 3\n",
|
| 1024 |
"y_i = X[..., 0] ** 2 + 6 * np.cos(2 * X[..., 2])\n",
|
| 1025 |
"y = np.sum(y_i, axis=1) / y_i.shape[1]\n",
|
|
|
|
| 1300 |
]
|
| 1301 |
},
|
| 1302 |
{
|
| 1303 |
+
"attachments": {},
|
| 1304 |
"cell_type": "markdown",
|
| 1305 |
"metadata": {
|
| 1306 |
"id": "6WuaeqyqbDhe"
|
|
|
|
| 1308 |
"source": [
|
| 1309 |
"Recall we are searching for $y_i$ above:\n",
|
| 1310 |
"\n",
|
| 1311 |
+
"$$ z = y^2,\\quad y = \\frac{1}{10} \\sum(y_i),\\quad y_i = x_{i0}^2 + 6 \\cos(2 x_{i2})$$"
|
| 1312 |
]
|
| 1313 |
},
|
| 1314 |
{
|
|
|
|
| 1375 |
},
|
| 1376 |
"gpuClass": "standard",
|
| 1377 |
"kernelspec": {
|
| 1378 |
+
"display_name": "Python (main_ipynb)",
|
| 1379 |
+
"language": "python",
|
| 1380 |
+
"name": "main_ipynb"
|
| 1381 |
},
|
| 1382 |
"language_info": {
|
| 1383 |
+
"name": "python",
|
| 1384 |
+
"version": "3.10.9"
|
| 1385 |
}
|
| 1386 |
},
|
| 1387 |
"nbformat": 4,
|