No abstract

The complexity of the electroencephalogram (EEG) during human sleep can be estimated by calculating the correlation dimension. Due to the large number of calculations required by this approach, only selected short (4-164 s) segments of the sleep EEG have been analysed previously. By using a new type of personal supercomputer, we were able to calculate the correlation dimension of overlapping 1 min EEG segments for the entire sleep episode (480 min) of 11 subjects and thereby delineate the time course of the changes. The correlation dimension was high in episodes of rapid eye movement (REM) sleep, declined progressively within each non-REM sleep episode, and reached a low level at times when EEG slow waves (0.75-4.5 Hz) were dominant. However, whereas slow-wave activity showed its typical progressive decline from non-REM/REM sleep cycle 1 to 4, no such trend was present for the correlation dimension. By providing an estimate of the complexity of a signal and being independent of amplitude and frequency measures, the correlation dimension represents a novel approach to exploring the dynamics of sleep and the processes underlying its regulation.

This paper describes the implementation of a fast neural net simulator on a novel parallel distributed-memory computer. A 60-processor system, named MUSIC (multiprocessor system with intelligent communication), is operational and runs the backpropagation algorithm at a speed of 330 million connection updates per second (continuous weight update) using 32-b floating-point precision. This is equal to 1.4 Gflops sustained performance. The complete system with 3.8 Gflops peak performance consumes less than 800 W of electrical power and fits into a 19-in rack. While reaching the speed of modern supercomputers, MUSIC still can be used as a personal desktop computer at a researcher's own disposal. In neural net simulation, this gives a computing performance to a single user which was unthinkable before. The system's real-time interfaces make it especially useful for embedded applications.