Mathing with Theo

I’m currently doing some work on Mutli-Agent Algorithms for a company in the UK and I thought it might make a nice topic for a post.

A multi-agent algorithm is an algorithm that attempts to simulate a complex phenomenon and thereby solve or make useful predictions and strategies for the phenomenon using millions of simulated autonomous agents.

For example, consider the stock market. The price of each individual stock may seem quite random when viewed on a large scale. However, upon further inspection we can see that the current price of a stock is the result of thousands of interactions over time between a set of traders. Each trader brings his knowledge, his expectations and his goals to the trading floor causing small shifts in price.

(more…)

I used to work at DoubleClick… now Google… as a Statistical Analyst. In my short career there I learned that internet ads are very rarely clicked on. In fact a rate of 1 in 1000 clicks/deliveries would be considered pretty decent.

Lately I’ve noticed that life seems to be a bit like an Internet Advertisement.

(more…)

Back in high school we learned that any right triangle inscribed inside of a circle has vertices on a point on the circumference and two points on the opposite ends of the diameter (if the right angle vertex is on the circumference).

(more…)

One of the hard parts about life is choosing one option at the expense of another future opportunity

For instance. When looking for a job it’s rare that you get three competing offers at one time. It’s more likely that you can get one interview/offer this week, and a different possibly better offer next week. If we take the current possibly sub-optimal offer we forgo next weeks potential offer for the security of an immediate pay check. However if we don’t take today’s offer and wait for next week we may both have greater financial security and greater job satisfaction… or we may have nothing at all.

In Artificial Intelligence the same applies, every decision made is a decision that commits the program/robot to a course of action which forgoes future options.

The questions to ask are: How do I design a program/robot to make decisions when faced with choices? How do I allow the program to learn from mistakes and make generalizing principles to help guide it when faced with a decision next time?

Of course we might also ask ourselves these very same questions: How do I make decisions? What are my generalizing and guiding principles?

Mathematicians and Statisticians are often asked, and usually in a highly dubious tone, “so…. what are you going to do with that advanced stats/math degree?”. Sometimes it’s followed with, “teach?”

So many people must have asked this question that the department of Statistics at Colombia University put on a whole conference devoted to the subject. The conference included panelists from both Academia and Industry.

A Postdoc from the department, commented on the conference.

I thought the following was a good observation:

There is not as much flexibility in industry as with academia (research must be in the companies interests), however, the compensation is usually much better.

All industry panelists agreed that statisticians must be excited by data.

It’s true, to be a good mathematical/statistical researcher you must be excited by data!

Since greentheo.scroggles.com was heading in a more green/political/philosophical direction I thought I would start another blog mostly for sharing code snippets, discussing Artificial Intelligence, Machine Learning and other things Math related.

You might say that this is an Applied Math blog. (Yikes!)

So to kick it off here is an interesting article I found today about a company who provides a handwriting analysis service for potential employers.

(more…)