University of Birmingham > Talks@bham > Artificial Intelligence and Natural Computation seminars > Parameter tuning via abduction

Parameter tuning via abduction

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Hector Basevi.

In machine learning, and more generally in optimisation, the programmer needs to construct a model with a given mathematical or algorithmic structure but parameterised by constants which don’t have definitive known values. For example a linear regression has the form y=ax+b, but a and b are not initially known. Learning or optimisation techniques are used to find or improve these parameters. Note that the learning or optimisation algorithms themselves are parameterised by best-guess constants such as learning rate, mutation rate etc, which can be subjected to optimisation.

In this talk I will present joint work with my students Koko Muroya and Steven Cheung regarding a logical analysis of parameters (provisional constants) via abductive reasoning. Using the well established correspondence between logic and functional programming we developed a new programming language which handle such parameters in an ergonomic way, resulting in a language with good formal properties.

This is preliminary work. My aim is to collect views, suggestions and criticism from the AI group which will help us more realistically address real needs of ML and optimisation programming.

This talk is part of the Artificial Intelligence and Natural Computation seminars series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.


Talks@bham, University of Birmingham. Contact Us | Help and Documentation | Privacy and Publicity.
talks@bham is based on from the University of Cambridge.