Using Locally Weighted Learning for Modeling

At this point we take a break from examples using Vizier and discuss typical modeling applications where memory based learning can be useful. We will also discuss the pros and cons of memory based learning compared to neural networks, which are another commonly used function approximation tool.

• ``Hands-off'' non-parametric relation finding
• Low dimensional supervised learning
• Complex function of a subset of inputs
• Simple function of most inputs, but complex function of a few
• Complex function of a few features of many inputs
• Pros and cons vs. neural networks
• When should locally weighted learning be used?