Bayesian Regression

Regression is one of the most common and basic supervised learning tasks in machine learning. Suppose we’re given a dataset \(\mathcal{D}\) of the form

\[\mathcal{D} = \{ (X_i, y_i) \} \qquad \text{for}\qquad i=1,2,...,N\]

The goal of linear regression is to fit a function to the data of the form:

\[y = w X + b + \epsilon\]

where \(w\) and \(b\) are learnable parameters and \(\epsilon\) represents observation noise. Specifically \(w\) is a matrix of weights and \(b\) is a bias vector.

Let’s first implement linear regression in PyTorch and learn point estimates for the parameters \(w\) and \(b\). Then we’ll see how to incorporate uncertainty into our estimates by using Pyro to implement Bayesian linear regression.


As always, let’s begin by importing the modules we’ll need.

In [ ]:
import os
import numpy as np
import torch
import torch.nn as nn

import pyro
from pyro.distributions import Normal
from pyro.infer import SVI, Trace_ELBO
from pyro.optim import Adam
# for CI testing
smoke_test = ('CI' in os.environ)


We’ll generate a toy dataset with one feature and \(w = 3\) and \(b = 1\) as follows:

In [ ]:
N = 100  # size of toy data

def build_linear_dataset(N, p=1, noise_std=0.01):
    X = np.random.rand(N, p)
    # w = 3
    w = 3 * np.ones(p)
    # b = 1
    y = np.matmul(X, w) + np.repeat(1, N) + np.random.normal(0, noise_std, size=N)
    y = y.reshape(N, 1)
    X, y = torch.tensor(X).type(torch.Tensor), torch.tensor(y).type(torch.Tensor)
    data =, y), 1)
    assert data.shape == (N, p + 1)
    return data

Note that we generate the data with a fixed observation noise \(\sigma = 0.1\).


Now let’s define our regression model. We’ll use PyTorch’s nn.Module for this. Our input \(X\) is a matrix of size \(N \times p\) and our output \(y\) is a vector of size \(p \times 1\). The function nn.Linear(p, 1) defines a linear transformation of the form \(Xw + b\) where \(w\) is the weight matrix and \(b\) is the additive bias. As you can see, we can easily make this a logistic regression by adding a non-linearity in the forward() method.

In [ ]:
class RegressionModel(nn.Module):
    def __init__(self, p):
        # p = number of features
        super(RegressionModel, self).__init__()
        self.linear = nn.Linear(p, 1)

    def forward(self, x):
        return self.linear(x)

regression_model = RegressionModel(1)


We will use the mean squared error (MSE) as our loss and Adam as our optimizer. We would like to optimize the parameters of the regression_model neural net above. We will use a somewhat large learning rate of 0.01 and run for 500 iterations.

In [ ]:
loss_fn = torch.nn.MSELoss(size_average=False)
optim = torch.optim.Adam(regression_model.parameters(), lr=0.05)
num_iterations = 1000 if not smoke_test else 2

def main():
    data = build_linear_dataset(N)
    x_data = data[:, :-1]
    y_data = data[:, -1]
    for j in range(num_iterations):
        # run the model forward on the data
        y_pred = regression_model(x_data).squeeze(-1)
        # calculate the mse loss
        loss = loss_fn(y_pred, y_data)
        # initialize gradients to zero
        # backpropagate
        # take a gradient step
        if (j + 1) % 50 == 0:
            print("[iteration %04d] loss: %.4f" % (j + 1, loss.item()))
    # Inspect learned parameters
    print("Learned parameters:")
    for name, param in regression_model.named_parameters():
        print("%s: %.3f" % (name,

if __name__ == '__main__':

Sample Output:

[iteration 0400] loss: 0.0105
[iteration 0450] loss: 0.0096
[iteration 0500] loss: 0.0095
[iteration 0550] loss: 0.0095
[iteration 0600] loss: 0.0095
[iteration 0650] loss: 0.0095
[iteration 0700] loss: 0.0095
[iteration 0750] loss: 0.0095
[iteration 0800] loss: 0.0095
[iteration 0850] loss: 0.0095
[iteration 0900] loss: 0.0095
[iteration 0950] loss: 0.0095
[iteration 1000] loss: 0.0095
Learned parameters:
linear.weight: 3.004
linear.bias: 0.997

Not too bad - you can see that the regressor learned parameters that were pretty close to the ground truth of \(w = 3, b = 1\). But how confident should we be in these point estimates?

Bayesian modeling offers a systematic framework for reasoning about model uncertainty. Instead of just learning point estimates, we’re going to learn a distribution over values of the parameters \(w\) and \(b\) that are consistent with the observed data.

Bayesian Regression

In order to make our linear regression Bayesian, we need to put priors on the parameters \(w\) and \(b\). These are distributions that represent our prior belief about reasonable values for \(w\) and \(b\) (before observing any data).


In order to do this, we’ll ‘lift’ the parameters \(w\) and \(b\) to random variables. We can do this in Pyro via random_module(), which effectively takes a given nn.Module and turns it into a distribution over the same module; in our case, this will be a distribution over regressors. Specifically, each parameter in the original regression model is sampled from the provided prior. This allows us to repurpose vanilla regression models for use in the Bayesian setting. For example:

In [ ]:
loc = torch.zeros(1, 1)
scale = torch.ones(1, 1)
# define a unit normal prior
prior = Normal(loc, scale)
# overload the parameters in the regression module with samples from the prior
lifted_module = pyro.random_module("regression_module", regression_model, prior)
# sample a regressor from the prior
sampled_reg_model = lifted_module()


We now have all the ingredients needed to specify our model. First we define priors over \(w\) and \(b\). Because we’re uncertain about the parameters a priori, we’ll use relatively wide priors \(\mathcal{N}(\mu = 0, \sigma = 10)\). Then we wrap regression_model with random_module and sample an instance of the regressor, lifted_reg_model. We then run the regressor forward on the inputs x_data. Finally we use the obs argument to the pyro.sample statement to condition on the observed data y_data. Note that we use the same fixed observation noise that was used to generate the data.

In [ ]:
def model(data):
    # Create unit normal priors over the parameters
    loc, scale = torch.zeros(1, 1), 10 * torch.ones(1, 1)
    bias_loc, bias_scale = torch.zeros(1), 10 * torch.ones(1)
    w_prior = Normal(loc, scale).independent(1)
    b_prior = Normal(bias_loc, bias_scale).independent(1)
    priors = {'linear.weight': w_prior, 'linear.bias': b_prior}
    # lift module parameters to random variables sampled from the priors
    lifted_module = pyro.random_module("module", regression_model, priors)
    # sample a regressor (which also samples w and b)
    lifted_reg_model = lifted_module()
    with pyro.iarange("map", N):
        x_data = data[:, :-1]
        y_data = data[:, -1]

        # run the regressor forward conditioned on data
        prediction_mean = lifted_reg_model(x_data).squeeze(-1)
        # condition on the observed data
                    Normal(prediction_mean, 0.1 * torch.ones(data.size(0))),


In order to do inference we’re going to need a guide, i.e. a parameterized family of distributions over \(w\) and \(b\). Writing down a guide will proceed in close analogy to the construction of our model, with the key difference that the guide parameters need to be trainable. To do this we register the guide parameters in the ParamStore using pyro.param().

In [ ]:
softplus = torch.nn.Softplus()

def guide(data):
    # define our variational parameters
    w_loc = torch.randn(1, 1)
    # note that we initialize our scales to be pretty narrow
    w_log_sig = torch.tensor(-3.0 * torch.ones(1, 1) + 0.05 * torch.randn(1, 1))
    b_loc = torch.randn(1)
    b_log_sig = torch.tensor(-3.0 * torch.ones(1) + 0.05 * torch.randn(1))
    # register learnable params in the param store
    mw_param = pyro.param("guide_mean_weight", w_loc)
    sw_param = softplus(pyro.param("guide_log_scale_weight", w_log_sig))
    mb_param = pyro.param("guide_mean_bias", b_loc)
    sb_param = softplus(pyro.param("guide_log_scale_bias", b_log_sig))
    # guide distributions for w and b
    w_dist = Normal(mw_param, sw_param).independent(1)
    b_dist = Normal(mb_param, sb_param).independent(1)
    dists = {'linear.weight': w_dist, 'linear.bias': b_dist}
    # overload the parameters in the module with random samples
    # from the guide distributions
    lifted_module = pyro.random_module("module", regression_model, dists)
    # sample a regressor (which also samples w and b)
    return lifted_module()

Note that we choose Gaussians for both guide distributions. Also, to ensure positivity, we pass each log scale through a softplus() transformation (an alternative to ensure positivity would be an exp()-transformation).


To do inference we’ll use stochastic variational inference (SVI) (for an introduction to SVI, see SVI Part I). Just like in the non-Bayesian linear regression, each iteration of our training loop will take a gradient step, with the difference that in this case, we’ll use the ELBO objective instead of the MSE loss by constructing a Trace_ELBO object that we pass to SVI.

In [ ]:
optim = Adam({"lr": 0.05})
svi = SVI(model, guide, optim, loss=Trace_ELBO())

Here Adam is a thin wrapper around torch.optim.Adam (see here for a discussion). The complete training loop is as follows:

In [ ]:
def main():
    data = build_linear_dataset(N)
    for j in range(num_iterations):
        # calculate the loss and take a gradient step
        loss = svi.step(data)
        if j % 100 == 0:
            print("[iteration %04d] loss: %.4f" % (j + 1, loss / float(N)))

if __name__ == '__main__':

To take an ELBO gradient step we simply call the step method of SVI. Notice that the data argument we pass to step will be passed to both model() and guide().

Validating Results

Let’s compare the variational parameters we learned to our previous result:

In [ ]:
for name in pyro.get_param_store().get_all_param_names():
    print("[%s]: %.3f" % (name, pyro.param(name).data.numpy()))

Sample Output:

[guide_log_scale_weight]: -3.217
[guide_log_scale_bias]: -3.164
[guide_mean_weight]: 2.966
[guide_mean_bias]: 0.941

As you can see, the means of our parameter estimates are pretty close to the values we previously learned. Now, however, instead of just point estimates, the parameters guide_log_scale_weight and guide_log_scale_bias provide us with uncertainty estimates. (Note that the scales are in log-space here, so the more negative the value, the narrower the width).

Finally, let’s evaluate our model by checking its predictive accuracy on new test data. This is known as point evaluation. We’ll sample 20 neural nets from our posterior, run them on the test data, then average across their predictions and calculate the MSE of the predicted values compared to the ground truth.

In [ ]:
X = np.linspace(6, 7, num=20)
y = 3 * X + 1
X, y = X.reshape((20, 1)), y.reshape((20, 1))
x_data, y_data = torch.tensor(X).type(torch.Tensor), torch.tensor(y).type(torch.Tensor)
loss = nn.MSELoss()
y_preds = torch.zeros(20, 1)
for i in range(20):
    # guide does not require the data
    sampled_reg_model = guide(None)
    # run the regression model and add prediction to total
    y_preds = y_preds + sampled_reg_model(x_data)
# take the average of the predictions
y_preds = y_preds / 20
print("Loss: ", loss(y_preds, y_data).item())

Sample Output:

Loss:  0.00025596367777325213

See the full code on Github.