|a.|Suppose you want to predict whether a human brain MRI has disease or not. Suppose you have m number of training samples, with each training sample having six extracted image features of mean, variance, skewness, kurtosis, entropy, and energy along with the class label (1 or 0 for disease or no disease). How can you solve this problem? Explain stepwise by giving all the mathematical and algorithmic details.|

|b.|Suppose after deploying the above model, you find out that the above model is too simple, and that the actual decision boundary is non-linear. How can you address this issue. You should also discuss the mathematical and algorithmic changes that may be required in the original model. |