liveBook · Manning

Kernel function

As explained in Build AI Drug Discovery Pipelines

A kernel function is a smooth, symmetric, non-negative weighting function that determines how an observation contributes to an estimated density at a location.

A kernel function is a smooth, symmetric, non-negative weighting function used by kernel density estimation (KDE). Centered on each observation, it gives nearby points more influence than distant points when estimating probability density. section 7.3.3

In a density estimate, the kernel:

  • Shapes each observation’s contribution to the estimated distribution. section 7.3.3
  • Supports continuous estimates that avoid histogram interval artifacts. section 7.3.3

Pharmacophore pair distributions

A Gaussian kernel estimates distributions of pairwise pharmacophore distances from fragment ensembles. Grid search selects the bandwidth while fitting scikit-learn KDE models.

How Build AI Drug Discovery Pipelines covers it

How deep it goes
Defines kernel shape and influence, then applies Gaussian KDE with cross-validated bandwidth selection.
Examples
Uses Gaussian and cosine kernels, bimodal data, and fragment-ensemble pharmacophore pair distances.
What it assumes
Assumes observations and density estimation, while defining kernel properties and influence from first principles.
What you can do after
Apply it via scikit-learn to fit KDE models and estimate pharmacophore distance distributions.
How it connects
Connects kernel functions to kernel density estimation, bandwidth, histograms, probability density, and pharmacophore distances.

Also explained in

Compare how 6 books explain kernel function →

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Build AI Drug Discovery Pipelines. The section links point to the parts of the book it draws on.