| Parallel Cascade Identification (PCI) (Korenberg, 1991) is an iterative algorithm that represents nonlinear systems by assembling parallel paths of cascades, each of which consists of a dynamic linear element followed by a static nonlinear element. This algorithm is based on a Volterra series expansion of a function that represents the unknown system. Knowing the stimulus (input) and response (output) of a system, we can use PCI to help us identify and model the dynamics of various systems. We have studied two biological systems: intracellular signal detection (in Chlamydomonas reinhardtii) and gesture recognition.;For intracellular signal detection, a key feature in our approach is the use of multiple inputs with different dynamical rates. Since experimental data are not available yet, simulated results are shown.;For gesture recognition, parallel cascade has been modified to function as a classifier (Korenberg and Morin, 1997). We have adapted the parallel cascade to work in a real-time setting. Our experimental results on the gesture recognition project shows promise.;We conclude with a discussion on possible applications and open problems. |
