Bayesian Linear Regression

Implement a Bayesian linear regressor on PyTorch. Apply a normal prior on regressor weights with zero mean and a predetermined variance identical for all dimensions (isotropic covariance). Implement all algebraic equations required for calculating the posterior distribution explicitly (e.g. using [login to view URL] function, and such). You are not allowed to use any service from the [login to view URL] library. Test your model on the Boston Housing data set. Use a randomly chosen 90% of the instances for training and the remaining 10% for testing. Repeat this operation 10 times and print out the average root mean square error (RMSE). Submit a Python source code that implements both Bayesian linear regression and the testing scheme described above. The code should only print out the average RMSE to the console. The firm submission deadline is 14 March 2018, 23:55 Istanbul time. Only submissions uploaded on LMS will be counted as valid. Submissions made via email will not be graded. Teamwork and group submissions of any sort are strictly disallowed. Your submission has to be your individual and original work. Submitting any code piece taken from an external source (web or other) is strictly disallowed. Such an attempt will result in a 0 grade from this assignment if the external source is properly cited and an F grade from the source if not cited (which will be treated as a plagiarism attempt).