Learning Techniques in Planning

Lectures delivered at the 2006 Machine Learning Summer School, Canberra, Feb 2006

For a more up-to-date version, check out the ICAPS 2007 Tutorial


In these lectures, I aim to provide an overview of the learning techniques that have found use in automated planning. Unlike most the clustering and classification tasks that have dominated the recent machine learning literature, learning in planning requires handling relational and first order representations, and foregrounds the need for knowledge-intensive learning techniques.

I will start with a brief review of the planning models, and discuss the opportunities for learning in planning. I will then provide a survey of the explanation-based, case-based and inductive learning techniques that have been successfully used to tackle them.

Slides (final version; as delivered):

Audio of lecture 1 [Feb 14 , 2006] (1hr 45min; 25MB)

Audio of lecture 1 [Feb 15, 2006] (1hr 18min; 18MB)

(Video of the lectures) Learning techniques in Planning

Rao Kambhampati
4 videos


Subbarao Kambhampati
Last modified: Thu Sep 27 12:47:04 MST 2007