Skip to content

Instantly share code, notes, and snippets.

Show Gist options
  • Save bcolloran/d57476f20e9944deae5f54f786649ad9 to your computer and use it in GitHub Desktop.
Save bcolloran/d57476f20e9944deae5f54f786649ad9 to your computer and use it in GitHub Desktop.
{
"cells": [
{
"cell_type": "markdown",
"source": [
"# Reproducing Sklearn results with Gpytorch #6\n",
"## (now using LBFGS optimizer from Botorch -- success!)\n",
"\n",
"This is the latest in a series of notebooks in which I've been attempting to reproduce the quality of the fits I'm able to get from sklearn when using gpytorch.\n",
"\n",
"For context, see previous notebooks:\n",
"* [version 4](https://gist.github.com/bcolloran/46de7c2aafef2fcd6c9e45c59f9b016b) - First attempt to use Pytorch's version of LBFGS (which is what sklearn uses) for optimization.\n",
"* [version 5](https://gist.github.com/bcolloran/9b450cb8bd02b25fd1a45314b4a833b9#file-simple_gp_regression-periodic-sklearn-mauna-rev5-lbfgs-zeromean-ipynb) - Switched to ZeroMean [per gpleiss's suggestion](https://github.com/cornellius-gp/gpytorch/discussions/1671#discussioncomment-966583), and standardized the Y values to be compatible with the zero mean constraint.\n",
"* **THIS VERSION** (version 6) - per [wjmaddox's suggestion](https://github.com/cornellius-gp/gpytorch/discussions/1671#discussioncomment-1034200) switched from Gpytorch's own LBFGS implementation to Botorch's `fit_gpytorch_model` (which uses scipy's L-BFGS-B implementation under the hood). Now getting pretty good fits! In the handful of runs I've done, I'm still not getting likelihoods quite as good as wjmaddox (or sklearn), but huge progress. I'll call it good enough!\n",
"\n",
"\n",
"-----\n",
"\n",
"### Outline:\n",
"* imports and loading data\n",
"* set up code for fitting sklearn models\n",
"* set up code for fitting gpytorch models\n",
"* define kernels with the same initialization\n",
"* verify that initialization is identitical\n",
"* compare results of optimizing gpytorch and sklearn models from same initialization"
],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 1,
"source": [
"import torch\n",
"import gpytorch\n",
"from gpytorch.kernels import Kernel, PeriodicKernel, RBFKernel, RQKernel, ScaleKernel, ProductKernel, AdditiveKernel\n",
"from matplotlib import pyplot as plt\n",
"from sklearn.preprocessing import StandardScaler\n",
"from sklearn.gaussian_process import GaussianProcessRegressor\n",
"from sklearn.gaussian_process.kernels \\\n",
" import RBF, WhiteKernel, RationalQuadratic, ExpSineSquared, ConstantKernel\n",
"import numpy as np\n",
"from typing import Union\n",
"from multimethod import multimethod\n",
"\n",
"from sklearn.datasets import fetch_openml\n",
"from botorch import fit_gpytorch_model\n",
"\n",
"%matplotlib inline"
],
"outputs": [],
"metadata": {}
},
{
"cell_type": "markdown",
"source": [
"### Load Mauna Loa data"
],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 2,
"source": [
"def load_mauna_loa_atmospheric_co2():\n",
" ml_data = fetch_openml(data_id=41187, as_frame=False)\n",
" months = []\n",
" ppmv_sums = []\n",
" counts = []\n",
"\n",
" y = ml_data.data[:, 0]\n",
" m = ml_data.data[:, 1]\n",
" month_float = y + (m - 1) / 12\n",
" ppmvs = ml_data.target\n",
"\n",
" for month, ppmv in zip(month_float, ppmvs):\n",
" if not months or month != months[-1]:\n",
" months.append(month)\n",
" ppmv_sums.append(ppmv)\n",
" counts.append(1)\n",
" else:\n",
" # aggregate monthly sum to produce average\n",
" ppmv_sums[-1] += ppmv\n",
" counts[-1] += 1\n",
" \n",
" months = np.array(months).reshape(-1, 1)\n",
" avg_ppmvs = np.array(ppmv_sums) / counts\n",
" return months, avg_ppmvs\n",
"\n",
"def to_torch_cuda(V):\n",
" return torch.tensor(V).flatten().double().cuda()\n",
"\n",
"X_raw, Y_raw = load_mauna_loa_atmospheric_co2()\n",
"\n",
"Y = StandardScaler().fit_transform(Y_raw.reshape(-1,1)).flatten()\n",
"\n",
"X_test = np.linspace(X_raw.min(), X_raw.max() + 30, 1000)[:, np.newaxis]\n",
"\n",
"f, ax = plt.subplots(1, 1, figsize=(14, 4))\n",
"ax.plot(X_raw, Y)\n",
"\n",
"X_torch = to_torch_cuda(X_raw)\n",
"Y_torch = to_torch_cuda(Y)\n",
"X_test_torch = to_torch_cuda(X_test)"
],
"outputs": [
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 1008x288 with 1 Axes>"
],
"image/png": "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"
},
"metadata": {
"needs_background": "light"
}
}
],
"metadata": {}
},
{
"cell_type": "markdown",
"source": [
"# Scikit learn code for fitting GPs"
],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 3,
"source": [
"def fit_sklearn_gp(kernel):\n",
" # SET `normalize_y=False` !!!\n",
" gp = GaussianProcessRegressor(kernel=kernel, alpha=0, normalize_y=False)\n",
" gp.fit(X_raw, Y)\n",
"\n",
" log_likelihood = gp.log_marginal_likelihood(gp.kernel_.theta)\n",
" # print(\"\\nLearned kernel:\\n%s\" % gp.kernel_)\n",
" # print(\"Log-marginal-likelihood: %.3f\" % log_likelihood)\n",
"\n",
" y_pred, y_std = gp.predict(X_test, return_std=True)\n",
"\n",
" # Illustration\n",
" f, ax = plt.subplots(1, 1, figsize=(14, 5))\n",
" ax.plot(X_raw, Y, 'k.', markersize=4)\n",
" ax.plot(X_test, y_pred,\"-\", color=\"steelblue\")\n",
" ax.fill_between(X_test[:, 0], y_pred - 2*y_std, y_pred + 2*y_std, alpha=0.3, color='steelblue')\n",
" ax.set_title(f\"log_likelihood: {log_likelihood:.2f}\")\n",
" return gp"
],
"outputs": [],
"metadata": {}
},
{
"cell_type": "markdown",
"source": [
"# Gpytorch code for fitting models"
],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 4,
"source": [
"# This block just contains function to print gpytorch kernel info\n",
"# in a format similar to sklearn, so that it's easy to check that kernels\n",
"# specs are the same\n",
"\n",
"def kernel_details_str(k: Kernel) -> str:\n",
" if isinstance(k, RBFKernel):\n",
" return f\"RBF(length_scale={k.lengthscale.item():.3f})\"\n",
"\n",
" elif isinstance(k, PeriodicKernel):\n",
" return f\"ExpSineSquared(length_scale={k.lengthscale.item():.3f}, periodcity={k.period_length.item():.3f},)\"\n",
"\n",
" elif isinstance(k, RQKernel):\n",
" return f\"RationalQuadratic(α={k.alpha.item():.3f},length_scale={k.lengthscale.item():.3f})\"\n",
"\n",
" elif isinstance(k, AdditiveKernel):\n",
" return f\"+\"\n",
"\n",
" elif isinstance(k, ProductKernel):\n",
" return f\"*\"\n",
"\n",
" else:\n",
" raise TypeError(\"Invalid kernel type\")\n",
"\n",
"@multimethod\n",
"def print_kernel(leaf: Union[RBFKernel,PeriodicKernel,RQKernel]) -> str:\n",
" return kernel_details_str(leaf)\n",
"\n",
"\n",
"@print_kernel.register\n",
"def print_kernel_scale(node: ScaleKernel) -> str:\n",
" scale = node.outputscale.item()\n",
" return f\"{np.sqrt(scale):.2f}**2 * {print_kernel(node.base_kernel)}\"\n",
"\n",
"\n",
"@print_kernel.register\n",
"def print_kernel_op_node(node: Union[AdditiveKernel, ProductKernel]) -> str:\n",
" children = f\" {kernel_details_str(node)} \".join([print_kernel(k) for k in node.kernels])\n",
" return f\"({children})\"\n",
"\n"
],
"outputs": [],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 5,
"source": [
"class ExactGPModel(gpytorch.models.ExactGP):\n",
" def __init__(self, train_x, train_y, likelihood, kernel):\n",
" super(ExactGPModel, self).__init__(train_x, train_y, likelihood)\n",
" self.mean_module = gpytorch.means.ZeroMean()\n",
" self.covar_module = kernel\n",
" \n",
" def forward(self, x):\n",
" mean_x = self.mean_module(x)\n",
" covar_x = self.covar_module(x)\n",
" return gpytorch.distributions.MultivariateNormal(mean_x, covar_x)\n",
"\n",
"\n",
"def log_marginal_likelihood(model, X, Y):\n",
" L = model.covar_module(X).detach().cholesky().clone().cpu().numpy()\n",
" Y = Y.clone().detach().cpu()\n",
" N = Y.shape[0]\n",
" sigma2 = model.likelihood.noise.item()\n",
"\n",
" K = L @ L.T + sigma2 * np.eye(N)\n",
" K_inv = np.linalg.inv(K)\n",
" _, log_det_K = np.linalg.slogdet(K)\n",
"\n",
" return -0.5 * (Y.T @ K_inv @ Y + log_det_K + N * np.log(2 * np.pi))\n",
"\n",
"\n",
"def plot_model(model, likelihood, train_x, train_y, test_x):\n",
"\n",
" model.eval()\n",
" likelihood.eval()\n",
"\n",
" # Make predictions by feeding model through likelihood\n",
" with torch.no_grad(), gpytorch.settings.fast_pred_var():\n",
"\n",
" observed_pred = likelihood(model(test_x))\n",
" # bigger plot of the last model\n",
" f, ax = plt.subplots(1, 1, figsize=(14, 5))\n",
" # Get upper and lower confidence bounds\n",
" lower, upper = observed_pred.confidence_region()\n",
" # Plot training data as black stars\n",
" ax.plot(train_x.cpu().numpy(), train_y.cpu().numpy(), 'k.', markersize=4)\n",
" # Plot predictive means as blue line\n",
" ax.plot(test_x.cpu().numpy(), observed_pred.mean.cpu().numpy(), \"-\", color=\"steelblue\")\n",
" # Shade between the lower and upper confidence bounds\n",
" ax.fill_between(test_x.cpu().numpy(), lower.cpu().numpy(), upper.cpu().numpy(), alpha=0.3)\n",
" log_likelihood = log_marginal_likelihood(model, train_x, train_y)\n",
"\n",
" ax.set_title(f\"log_likelihood: {log_likelihood:.2f}\")\n"
],
"outputs": [],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 6,
"source": [
"def create_and_train_model(\n",
" kernel_gen_fn,\n",
" optimization_fn,\n",
" train_x,\n",
" train_y,\n",
" test_x,\n",
" likelihood_noise = None,\n",
" training_iter = 100, \n",
"):\n",
" likelihood = gpytorch.likelihoods.GaussianLikelihood().cuda()\n",
" if likelihood_noise is not None:\n",
" likelihood.initialize(noise=likelihood_noise) \n",
" model = ExactGPModel(train_x, train_y, likelihood, kernel_gen_fn() ).cuda()\n",
" # \"Loss\" for GPs - the marginal log likelihood\n",
" mll = gpytorch.mlls.ExactMarginalLogLikelihood(likelihood, model)\n",
"\n",
" model.train()\n",
" likelihood.train()\n",
" losses = optimization_fn(\n",
" model,\n",
" mll,\n",
" train_x,\n",
" train_y,\n",
" training_iter, \n",
" )\n",
"\n",
" plot_model(model, likelihood, train_x, train_y, test_x)\n",
" return model, likelihood, losses\n",
"\n",
"\n",
"def optimize_LBFGS(\n",
" model,\n",
" mll,\n",
" train_x,\n",
" train_y,\n",
" training_iter, \n",
"):\n",
" fit_gpytorch_model(mll)\n",
" return None\n",
"\n",
"\n",
"def optimize_adam(\n",
" model,\n",
" mll,\n",
" train_x,\n",
" train_y,\n",
" training_iter, \n",
"): \n",
" optimizer = torch.optim.Adam(model.parameters(), lr=0.1) # Includes \n",
"\n",
" losses = []\n",
"\n",
" for i in range(training_iter):\n",
" # Zero gradients from previous iteration\n",
" optimizer.zero_grad()\n",
" # Output from model\n",
" output = model(train_x)\n",
" # Calc loss and backprop gradients\n",
" loss = -mll(output, train_y)\n",
" losses.append(loss.item())\n",
" loss.backward()\n",
" optimizer.step()\n",
"\n",
" return losses\n"
],
"outputs": [],
"metadata": {}
},
{
"cell_type": "markdown",
"source": [
"# Set up kernels\n",
"\n",
"In this section, we'll set up out kernels and make sure they look the same before the are optimized.\n",
"\n",
"Note that the sklearn `WhiteKernel` is equivalent to the `noise` in the likelihood of a gpytorch model.\n",
"\n",
"(See [this notebook](https://gist.github.com/bcolloran/7686b63a5544295f0382415e553194ea) for verification of this equivalence)"
],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 7,
"source": [
"# sklearn kernel definitions\n",
"def sklearn_default_kernel():\n",
" # Kernel with DEFAULT PARAMETERS - no scaling etc\n",
" mean_kernel = ConstantKernel(constant_value=0, constant_value_bounds=\"fixed\")\n",
" k1 = ConstantKernel() * RBF() # long term smooth rising trend\n",
" k2 = ConstantKernel() * RBF() * ExpSineSquared() # seasonal component\n",
"\n",
" # medium term irregularities\n",
" k3 = ConstantKernel() * RationalQuadratic()\n",
"\n",
" # noise terms -- we'll keep these\n",
" k4 = ConstantKernel() * RBF() + WhiteKernel(noise_level=0.1**2, noise_level_bounds=(1e-5, np.inf)) \n",
"\n",
" return mean_kernel + k1 + k2 + k3 + k4\n"
],
"outputs": [],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 8,
"source": [
"# NOTE: because of the gpytorch's constraint transforms,\n",
"# to match skearn's defaults we do actually have to define initial values\n",
"def gpytorch_default_kernel():\n",
" #### long term\n",
" k_long_term_RBF = RBFKernel()\n",
" k_long_term_RBF.lengthscale = 1\n",
"\n",
" k_long_term = ScaleKernel(k_long_term_RBF)\n",
" k_long_term.outputscale = 1\n",
"\n",
" #### seasonal\n",
" k_seasonal_RBF = RBFKernel()\n",
" k_seasonal_RBF.lengthscale = 1\n",
"\n",
" k_seasonal_periodic = PeriodicKernel()\n",
" k_seasonal_periodic.period_length = 1.0\n",
" # this must be squared, see https://github.com/cornellius-gp/gpytorch/issues/1020\n",
" k_seasonal_periodic.lengthscale = 1\n",
"\n",
" k_seasonal = ScaleKernel(k_seasonal_RBF * k_seasonal_periodic)\n",
" k_seasonal.outputscale = 1\n",
"\n",
" ### medium term\n",
" k_medium_RQ = RQKernel()\n",
" k_medium_RQ.alpha = 1.0\n",
" k_medium_RQ.lengthscale = 1.0\n",
"\n",
" k_medium = ScaleKernel(k_medium_RQ)\n",
" k_medium.outputscale = 1\n",
"\n",
" #### noise term\n",
" k_noise_RBF = RBFKernel()\n",
" k_noise_RBF.lengthscale = 1\n",
"\n",
" k_noise = ScaleKernel(k_noise_RBF)\n",
" k_noise.outputscale = 1\n",
"\n",
" return k_long_term + k_seasonal + k_medium + k_noise\n",
"\n",
" return k_long_term + k_seasonal + k_medium + k_noise"
],
"outputs": [],
"metadata": {}
},
{
"cell_type": "markdown",
"source": [
"## Verify that initial kernels are identical (or very close)"
],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 9,
"source": [
"# VERIFY THAT THESE KERNELS LOOK THE SAME BEFORE OPTIMIZATION!!!\n",
"print(\"--- SKLEARN KERNEL ---\")\n",
"print(sklearn_default_kernel(), \"\\n\")\n",
"\n",
"print(\"--- GPYTORCH KERNEL ---\")\n",
"print(print_kernel(gpytorch_default_kernel()), \"\\n\")\n"
],
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"--- SKLEARN KERNEL ---\n",
"0**2 + 1**2 * RBF(length_scale=1) + 1**2 * RBF(length_scale=1) * ExpSineSquared(length_scale=1, periodicity=1) + 1**2 * RationalQuadratic(alpha=1, length_scale=1) + 1**2 * RBF(length_scale=1) + WhiteKernel(noise_level=0.01) \n",
"\n",
"--- GPYTORCH KERNEL ---\n",
"(1.00**2 * RBF(length_scale=1.000) + 1.00**2 * (RBF(length_scale=1.000) * ExpSineSquared(length_scale=1.000, periodcity=1.000,)) + 1.00**2 * RationalQuadratic(α=1.000,length_scale=1.000) + 1.00**2 * RBF(length_scale=1.000)) \n",
"\n"
]
}
],
"metadata": {}
},
{
"cell_type": "markdown",
"source": [
"# Optimizing from identical initial kernels\n"
],
"metadata": {}
},
{
"cell_type": "markdown",
"source": [
"## Starting from DEFAULT kernel (sklearn, gpytorch (adam), gpytorch (LBFGS))"
],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 10,
"source": [
"gp = fit_sklearn_gp(sklearn_default_kernel())"
],
"outputs": [
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 1008x360 with 1 Axes>"
],
"image/png": "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"
},
"metadata": {
"needs_background": "light"
}
}
],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 11,
"source": [
"print(\"before optimization\")\n",
"print(gp.kernel)\n",
"print(\"after optimization\")\n",
"print(gp.kernel_)"
],
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"before optimization\n",
"0**2 + 1**2 * RBF(length_scale=1) + 1**2 * RBF(length_scale=1) * ExpSineSquared(length_scale=1, periodicity=1) + 1**2 * RationalQuadratic(alpha=1, length_scale=1) + 1**2 * RBF(length_scale=1) + WhiteKernel(noise_level=0.01)\n",
"after optimization\n",
"0**2 + 1.56**2 * RBF(length_scale=50.1) + 0.311**2 * RBF(length_scale=223) * ExpSineSquared(length_scale=1.81, periodicity=1) + 0.0804**2 * RationalQuadratic(alpha=0.0213, length_scale=0.887) + 1.56**2 * RBF(length_scale=50.1) + WhiteKernel(noise_level=0.000133)\n"
]
}
],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 12,
"source": [
"model, likelihood, losses = create_and_train_model(\n",
" kernel_gen_fn = gpytorch_default_kernel,\n",
" optimization_fn = optimize_adam,\n",
" train_x = X_torch,\n",
" train_y = Y_torch,\n",
" test_x = X_test_torch,\n",
" likelihood_noise = 0.1**2\n",
")\n",
"print(\n",
" print_kernel(model.covar_module), \"\\n\",\n",
" likelihood.noise.item()\n",
")"
],
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"(0.47**2 * RBF(length_scale=4.947) + 0.21**2 * (RBF(length_scale=0.715) * ExpSineSquared(length_scale=1.236, periodcity=1.523,)) + 0.39**2 * RationalQuadratic(α=3.356,length_scale=4.664) + 0.47**2 * RBF(length_scale=4.947)) \n",
" 0.00015337555669248104\n"
]
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 1008x360 with 1 Axes>"
],
"image/png": "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"
},
"metadata": {
"needs_background": "light"
}
}
],
"metadata": {}
},
{
"cell_type": "code",
"execution_count": 13,
"source": [
"model, likelihood, losses = create_and_train_model(\n",
" kernel_gen_fn = gpytorch_default_kernel,\n",
" optimization_fn = optimize_LBFGS,\n",
" train_x = X_torch,\n",
" train_y = Y_torch,\n",
" test_x = X_test_torch,\n",
" likelihood_noise = 0.1**2\n",
")\n",
"\n",
"print(\n",
" print_kernel(model.covar_module), \"\\n\",\n",
" likelihood.noise.item()\n",
")"
],
"outputs": [
{
"output_type": "stream",
"name": "stdout",
"text": [
"(0.02**2 * RBF(length_scale=0.583) + 0.10**2 * (RBF(length_scale=65.104) * ExpSineSquared(length_scale=0.310, periodcity=1.999,)) + 1.99**2 * RationalQuadratic(α=6.365,length_scale=35.787) + 0.02**2 * RBF(length_scale=0.583)) \n",
" 0.00018807227024808526\n"
]
},
{
"output_type": "stream",
"name": "stderr",
"text": [
"/home/bc/miniconda3/lib/python3.9/site-packages/gpytorch/utils/cholesky.py:44: NumericalWarning: A not p.d., added jitter of 1.0e-08 to the diagonal\n",
" warnings.warn(f\"A not p.d., added jitter of {jitter_new:.1e} to the diagonal\", NumericalWarning)\n"
]
},
{
"output_type": "display_data",
"data": {
"text/plain": [
"<Figure size 1008x360 with 1 Axes>"
],
"image/png": "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"
},
"metadata": {
"needs_background": "light"
}
}
],
"metadata": {}
}
],
"metadata": {
"anaconda-cloud": {},
"kernelspec": {
"name": "python3",
"display_name": "Python 3.9.1 64-bit ('base': conda)"
},
"language_info": {
"codemirror_mode": {
"name": "ipython",
"version": 3
},
"file_extension": ".py",
"mimetype": "text/x-python",
"name": "python",
"nbconvert_exporter": "python",
"pygments_lexer": "ipython3",
"version": "3.9.1"
},
"interpreter": {
"hash": "a625ebe7cc536cd8bdf30a2e298dba7ca2d4a19bf14b47d0bf6039ccbb3c856a"
}
},
"nbformat": 4,
"nbformat_minor": 1
}
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment