Contents
Do you need MLE to estimate Gaussian parameters?
To take the leap from data → model, we need to not only estimate possible parameters of the model (eg. for Gaussians we need [μ, Σ] ), but we want the best model possible for our data. That’s where MLE comes into play.
Which is a defining characteristic of MLE for Gaussians?
The defining characteristic of MLE is that it uses only existing data to estimate parameters of the model. This is in contrast to approaches which exploit prior knowledge in addition to existing data. 1 Today, we’re talking about MLE for Gaussians, so this is going to be a classification task.
How to calculate the likelihood of a Gaussian?
Likelihood for a Gaussian We assume the data we’re working with was generated by an underlying Gaussian process in the real world. As such, the likelihood function () is the Gaussian itself. Therefore, for MLE of a Gaussian model, we will need to find good estimates of both parameters: and :
Why do we want the best parameters for a Gaussian distribution?
That’s because the log() function is monotonically increasing, and therefore So now, we know that we want to get the best parameters θ = [μ, Σ] for a dataset X evaluating on a normal, Gaussian distribution.
Which is log marginal likelihood for Gaussian process?
Log marginal likelihood for Gaussian Process as per Rasmussen’s Gaussian Processes for Machine Learning equation 2.30 is: Implementations typically apply gradient descent by default to maximize this over some family of K, where that is typically a sum/product of constant kernels, RBF kernels, periodic kernels, etc
How to calculate the maximum likelihood of a model?
Write down a model for how we believe the data was generated. Set the parameters of our model to values which maximize the likelihood of the parameters given the data. We’re going to cover the basics of creating models, understanding likelihood, and using the Maximum Likelihood Estimation process for fitting our parameters.
How is maximum likelihood estimation used in data science?
Maximum Likelihood Estimation is a powerful technique for fitting our models to data. The solutions provided by MLE are often very intuitive, but they’re completely data driven. This means, that the more data we have, the more accurate our solutions become and vice versa.