A Bradley-Terry Artificial Neural Network Model for Individual ...
A Bradley-Terry Artificial Neural Network Model for Individual ...
A Bradley-Terry Artificial Neural Network Model for Individual ...
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
6 Joshua Menke, Tony Martinez<br />
thousands of comparisions and competitors. In addition, as mentioned, it is<br />
impossible to fit a model this way if new individuals can be added at any<br />
point. Elo and (Glickman, 1999) have both proposed methods <strong>for</strong> approximating<br />
a simultaneous fit by by adapting the <strong>Bradley</strong>-<strong>Terry</strong> model after<br />
every comparison. This also allows <strong>for</strong> new individuals to be inserted at any<br />
point in the model. However, neither of these methods has been extended to<br />
extract individual ratings from groups. The model presented here does allow<br />
individual ratings to be determined in addition to providing an efficient<br />
update after each comparison.<br />
3 The ANN <strong>Model</strong><br />
The section proceeds as follows: it first presents the basic framework <strong>for</strong><br />
viewing the <strong>Bradley</strong>-<strong>Terry</strong> model as a delta-rule trained single layer ANN<br />
in 3.1, then the next section, 3.2 will show how to extend the model to<br />
learn individual ratings within a group, 3.3 extends the model to allow<br />
different weights <strong>for</strong> each individual based on their contribution, 3.4 shows<br />
how the model can be extened to learn “home field” advantages, 3.5 presents<br />
a heuristic to deal with uncertainty in an individual’s rating, and 3.6 gives<br />
another heuristic <strong>for</strong> preventing rating inflation.<br />
3.1 The Basic <strong>Model</strong><br />
The basis <strong>for</strong> the <strong>Bradley</strong>-<strong>Terry</strong> ANN <strong>Model</strong> begins with the ANN shown<br />
in figure 1. Assuming, without loss of generality, that group A is always the