LAB REPORT

Simulation Yields Confusing Results

Polyorc Cognitive Computation reports difficulty
implementing code-optimizers in its neural
network AI project. The learning system's capacity
to write and rewrite its own code has "backfired
on the programmers", according to company officer and 
lead product developer Dwayne Olson.
   As the project approaches the limitations of
its hardware, code-optimization has become
essential if the experiment is to continue.
With the system's code so rigorously monitoring,
 interpreting, and drawing heuristic conclusions 
from every level of its operation, it has become 
impossible to troubleshoot.  Every change made
at one level of the code results in the program
shifting around the rest of its heirarchy, 
which effectively renders the original change meaningless, 
and presents the programmers with an entirely 
new set of problems to work through.
   Despite these setbacks, Polyorc's Heuristic Induction
Network has produced some interesting results,
including this vaguely representational image, "created"
seemingly by accident during the final phases of 
the system's visual-recognition coding:
 
Polyorc is currently seeking backing to conduct a full 
hardware replacement for the network. 

This page hosted by Get your own Free Home Page