A Model Free Method (MFM) is a strategy to solve some problem by avoiding, as far as possible, the use of Models of the problem domain. Models are simplifications of reality such as naïve models, scientific models, theories, hypotheses, formulas, equations, or computer programs. Over a dozen primitive MFMs have been identified and many others can be derived from those. The key criterion is whether we model anything in the problem domain or not. We're allowed to model other domains and meta-domains and still say we are using MFMs.

There are multiple reasons to use MFMs. I will start by discussing these reasons, quickly overview the primitive MFMs, and then we move to the actual Workshop part.

I'll present several problems, one by one, and we will solve each problem twice, at a high level, on the whiteboard. First we will solve it using Models; we'll discuss the limitations of each Model and discuss what might go wrong when using it. Then we'll solve the problem using MFMs, and will discuss and compare the merits of each solution and estimate the required effort and resulting precision for each.

As a warm-up we'll solve the problem of how to aim a cannon that was discussed in the presentation "Models vs Patterns" given a couple years ago and which is available as a video on http://videos.syntience.com . I'll be re-using those slides.

I know I'll include one problem from Genomics and we'll also discuss how to "get movies from the brain of what we see". I won't discuss the NetFlix challenge problem again since I've done it twice. If you can think of a good problem which showcases MFM solutions, send me mail.

Don't be late or you'll miss the most delightfully surprising intelligence-themed movie of the week.

I may or may not video record this meetup. No webcast.