diff --git a/examples/03_Multitask_Exact_GPs/Hadamard_Multitask_GP_Regression.ipynb b/examples/03_Multitask_Exact_GPs/Hadamard_Multitask_GP_Regression.ipynb index b49856625..37dd622eb 100644 --- a/examples/03_Multitask_Exact_GPs/Hadamard_Multitask_GP_Regression.ipynb +++ b/examples/03_Multitask_Exact_GPs/Hadamard_Multitask_GP_Regression.ipynb @@ -29,7 +29,22 @@ "cell_type": "code", "execution_count": 1, "metadata": {}, - "outputs": [], + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "[KeOps] Warning : \n", + " The default C++ compiler could not be found on your system.\n", + " You need to either define the CXX environment variable or a symlink to the g++ command.\n", + " For example if g++-8 is the command you can do\n", + " import os\n", + " os.environ['CXX'] = 'g++-8'\n", + " \n", + "[KeOps] Warning : Cuda libraries were not detected on the system or could not be loaded ; using cpu only mode\n" + ] + } + ], "source": [ "import math\n", "import torch\n", @@ -58,10 +73,10 @@ "metadata": {}, "outputs": [], "source": [ - "train_x1 = torch.rand(2000)\n", - "train_x2 = torch.rand(2000)\n", + "train_x1 = torch.rand(50)\n", + "train_x2 = torch.rand(50)\n", "\n", - "train_y1 = torch.sin(train_x1 * (2 * math.pi)) + torch.randn(train_x1.size()) * 0.2\n", + "train_y1 = torch.sin(train_x1 * (2 * math.pi)) + torch.randn(train_x1.size()) * 0.6\n", "train_y2 = torch.cos(train_x2 * (2 * math.pi)) + torch.randn(train_x2.size()) * 0.2" ] }, @@ -108,7 +123,7 @@ " \n", " return gpytorch.distributions.MultivariateNormal(mean_x, covar)\n", "\n", - "likelihood = gpytorch.likelihoods.GaussianLikelihood()\n", + "likelihood = gpytorch.likelihoods.HadamardGaussianLikelihood(num_tasks=2)\n", "\n", "train_i_task1 = torch.full((train_x1.shape[0],1), dtype=torch.long, fill_value=0)\n", "train_i_task2 = torch.full((train_x2.shape[0],1), dtype=torch.long, fill_value=1)\n", @@ -139,60 +154,68 @@ "scrolled": false }, "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "c:\\Users\\tobyb\\phd\\venvs\\gpytorchvenv\\lib\\site-packages\\linear_operator\\utils\\interpolation.py:71: UserWarning: torch.sparse.SparseTensor(indices, values, shape, *, device=) is deprecated. Please use torch.sparse_coo_tensor(indices, values, shape, dtype=, device=). (Triggered internally at ..\\torch\\csrc\\utils\\tensor_new.cpp:620.)\n", + " summing_matrix = cls(summing_matrix_indices, summing_matrix_values, size)\n" + ] + }, { "name": "stdout", "output_type": "stream", "text": [ - "Iter 1/50 - Loss: 1.000\n", - "Iter 2/50 - Loss: 0.960\n", - "Iter 3/50 - Loss: 0.918\n", - "Iter 4/50 - Loss: 0.874\n", - "Iter 5/50 - Loss: 0.831\n", - "Iter 6/50 - Loss: 0.788\n", - "Iter 7/50 - Loss: 0.745\n", - "Iter 8/50 - Loss: 0.702\n", - "Iter 9/50 - Loss: 0.659\n", - "Iter 10/50 - Loss: 0.618\n", - "Iter 11/50 - Loss: 0.580\n", - "Iter 12/50 - Loss: 0.545\n", - "Iter 13/50 - Loss: 0.511\n", - "Iter 14/50 - Loss: 0.478\n", - "Iter 15/50 - Loss: 0.445\n", - "Iter 16/50 - Loss: 0.412\n", - "Iter 17/50 - Loss: 0.378\n", - "Iter 18/50 - Loss: 0.344\n", - "Iter 19/50 - Loss: 0.309\n", - "Iter 20/50 - Loss: 0.276\n", - "Iter 21/50 - Loss: 0.243\n", - "Iter 22/50 - Loss: 0.211\n", - "Iter 23/50 - Loss: 0.182\n", - "Iter 24/50 - Loss: 0.154\n", - "Iter 25/50 - Loss: 0.129\n", - "Iter 26/50 - Loss: 0.105\n", - "Iter 27/50 - Loss: 0.085\n", - "Iter 28/50 - Loss: 0.067\n", - "Iter 29/50 - Loss: 0.052\n", - "Iter 30/50 - Loss: 0.039\n", - "Iter 31/50 - Loss: 0.029\n", - "Iter 32/50 - Loss: 0.021\n", - "Iter 33/50 - Loss: 0.015\n", - "Iter 34/50 - Loss: 0.012\n", - "Iter 35/50 - Loss: 0.010\n", - "Iter 36/50 - Loss: 0.009\n", - "Iter 37/50 - Loss: 0.009\n", - "Iter 38/50 - Loss: 0.009\n", - "Iter 39/50 - Loss: 0.009\n", - "Iter 40/50 - Loss: 0.008\n", - "Iter 41/50 - Loss: 0.008\n", - "Iter 42/50 - Loss: 0.009\n", - "Iter 43/50 - Loss: 0.009\n", - "Iter 44/50 - Loss: 0.010\n", - "Iter 45/50 - Loss: 0.010\n", - "Iter 46/50 - Loss: 0.010\n", - "Iter 47/50 - Loss: 0.010\n", - "Iter 48/50 - Loss: 0.008\n", - "Iter 49/50 - Loss: 0.007\n", - "Iter 50/50 - Loss: 0.005\n" + "Iter 1/50 - Loss: 1.118\n", + "Iter 2/50 - Loss: 1.084\n", + "Iter 3/50 - Loss: 1.047\n", + "Iter 4/50 - Loss: 1.010\n", + "Iter 5/50 - Loss: 0.975\n", + "Iter 6/50 - Loss: 0.941\n", + "Iter 7/50 - Loss: 0.908\n", + "Iter 8/50 - Loss: 0.876\n", + "Iter 9/50 - Loss: 0.847\n", + "Iter 10/50 - Loss: 0.823\n", + "Iter 11/50 - Loss: 0.803\n", + "Iter 12/50 - Loss: 0.787\n", + "Iter 13/50 - Loss: 0.774\n", + "Iter 14/50 - Loss: 0.761\n", + "Iter 15/50 - Loss: 0.748\n", + "Iter 16/50 - Loss: 0.734\n", + "Iter 17/50 - Loss: 0.720\n", + "Iter 18/50 - Loss: 0.704\n", + "Iter 19/50 - Loss: 0.689\n", + "Iter 20/50 - Loss: 0.673\n", + "Iter 21/50 - Loss: 0.658\n", + "Iter 22/50 - Loss: 0.643\n", + "Iter 23/50 - Loss: 0.629\n", + "Iter 24/50 - Loss: 0.616\n", + "Iter 25/50 - Loss: 0.604\n", + "Iter 26/50 - Loss: 0.593\n", + "Iter 27/50 - Loss: 0.584\n", + "Iter 28/50 - Loss: 0.576\n", + "Iter 29/50 - Loss: 0.571\n", + "Iter 30/50 - Loss: 0.567\n", + "Iter 31/50 - Loss: 0.566\n", + "Iter 32/50 - Loss: 0.567\n", + "Iter 33/50 - Loss: 0.568\n", + "Iter 34/50 - Loss: 0.570\n", + "Iter 35/50 - Loss: 0.572\n", + "Iter 36/50 - Loss: 0.572\n", + "Iter 37/50 - Loss: 0.573\n", + "Iter 38/50 - Loss: 0.573\n", + "Iter 39/50 - Loss: 0.573\n", + "Iter 40/50 - Loss: 0.574\n", + "Iter 41/50 - Loss: 0.574\n", + "Iter 42/50 - Loss: 0.574\n", + "Iter 43/50 - Loss: 0.574\n", + "Iter 44/50 - Loss: 0.573\n", + "Iter 45/50 - Loss: 0.572\n", + "Iter 46/50 - Loss: 0.571\n", + "Iter 47/50 - Loss: 0.570\n", + "Iter 48/50 - Loss: 0.568\n", + "Iter 49/50 - Loss: 0.567\n", + "Iter 50/50 - Loss: 0.566\n" ] } ], @@ -216,7 +239,7 @@ "for i in range(training_iterations):\n", " optimizer.zero_grad()\n", " output = model(full_train_x, full_train_i)\n", - " loss = -mll(output, full_train_y)\n", + " loss = -mll(output, full_train_y, [full_train_i])\n", " loss.backward()\n", " print('Iter %d/50 - Loss: %.3f' % (i + 1, loss.item()))\n", " optimizer.step()" @@ -236,14 +259,12 @@ "outputs": [ { "data": { - "image/png": "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\n", + "image/png": "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", "text/plain": [ - "
" + "
" ] }, - "metadata": { - "needs_background": "light" - }, + "metadata": {}, "output_type": "display_data" } ], @@ -266,8 +287,8 @@ "# The gpytorch.settings.fast_pred_var flag activates LOVE (for fast variances)\n", "# See https://arxiv.org/abs/1803.06058\n", "with torch.no_grad(), gpytorch.settings.fast_pred_var():\n", - " observed_pred_y1 = likelihood(model(test_x, test_i_task1))\n", - " observed_pred_y2 = likelihood(model(test_x, test_i_task2))\n", + " observed_pred_y1 = likelihood(model(test_x, test_i_task1), [test_i_task1])\n", + " observed_pred_y2 = likelihood(model(test_x, test_i_task2), [test_i_task2])\n", "\n", "# Define plotting function\n", "def ax_plot(ax, train_y, train_x, rand_var, title):\n", @@ -310,7 +331,7 @@ "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython3", - "version": "3.8.0" + "version": "3.10.11" } }, "nbformat": 4, diff --git a/gpytorch/likelihoods/__init__.py b/gpytorch/likelihoods/__init__.py index 31a370079..c48400312 100644 --- a/gpytorch/likelihoods/__init__.py +++ b/gpytorch/likelihoods/__init__.py @@ -9,6 +9,7 @@ GaussianLikelihood, GaussianLikelihoodWithMissingObs, ) +from .hadamard_gaussian_likelihood import HadamardGaussianLikelihood from .laplace_likelihood import LaplaceLikelihood from .likelihood import _OneDimensionalLikelihood, Likelihood from .likelihood_list import LikelihoodList @@ -27,6 +28,7 @@ "FixedNoiseGaussianLikelihood", "GaussianLikelihood", "GaussianLikelihoodWithMissingObs", + "HadamardGaussianLikelihood", "HeteroskedasticNoise", "LaplaceLikelihood", "Likelihood", diff --git a/gpytorch/likelihoods/hadamard_gaussian_likelihood.py b/gpytorch/likelihoods/hadamard_gaussian_likelihood.py new file mode 100644 index 000000000..99dc4b2f7 --- /dev/null +++ b/gpytorch/likelihoods/hadamard_gaussian_likelihood.py @@ -0,0 +1,81 @@ +#!/usr/bin/env python3 + +from typing import Any + +import torch + +from linear_operator.operators import DiagLinearOperator + +from ..distributions import base_distributions, MultivariateNormal +from ..likelihoods import _GaussianLikelihoodBase +from .noise_models import MultitaskHomoskedasticNoise + + +class HadamardGaussianLikelihood(_GaussianLikelihoodBase): + r""" + Likelihood for input-wise homo-skedastic noise, and task-wise + hetero-skedastic, i.e. we learn a different (constant) noise level for each fidelity. + + Args: + num_of_tasks : number of tasks in the multi output GP + noise_prior : any prior you want to put on the noise + noise_constraint : constraint to put on the noise + """ + + def __init__( + self, + num_tasks, + noise_prior=None, + noise_constraint=None, + batch_shape=torch.Size(), + **kwargs, + ): + noise_covar = MultitaskHomoskedasticNoise( + num_tasks=num_tasks, + noise_prior=noise_prior, + noise_constraint=noise_constraint, + batch_shape=batch_shape, + ) + self.num_tasks = num_tasks + super().__init__(noise_covar=noise_covar) + + @property + def noise(self) -> torch.Tensor: + return self.noise_covar.noise + + @noise.setter + def noise(self, value: torch.Tensor) -> None: + self.noise_covar.initialize(noise=value) + + @property + def raw_noise(self) -> torch.Tensor: + return self.noise_covar.raw_noise + + @raw_noise.setter + def raw_noise(self, value: torch.Tensor) -> None: + self.noise_covar.initialize(raw_noise=value) + + def _shaped_noise_covar(self, base_shape: torch.Size, *params: Any, **kwargs: Any): + # params contains task indexes + task_idxs = params[0][-1] + noise_base_covar_matrix = self.noise_covar(*params, shape=base_shape, **kwargs) + + all_tasks = torch.arange(self.num_tasks)[:, None] + diag = torch.eq(all_tasks, task_idxs.mT) + mask = DiagLinearOperator(diag) + return (noise_base_covar_matrix @ mask).sum(dim=-3) + + def forward( + self, + function_samples: torch.Tensor, + *params: Any, + **kwargs: Any, + ) -> base_distributions.Normal: + noise = self._shaped_noise_covar(function_samples.shape, *params, **kwargs).diag() + return base_distributions.Normal(function_samples, noise.sqrt()) + + def marginal(self, function_dist: MultivariateNormal, *params: Any, **kwargs: Any) -> MultivariateNormal: + mean, covar = function_dist.mean, function_dist.lazy_covariance_matrix + noise_covar = self._shaped_noise_covar(mean.shape, *params, **kwargs).squeeze(0) + full_covar = covar + noise_covar + return function_dist.__class__(mean, full_covar)