DYNAMIC APERIODIC NEURAL NETWORK FOR TIME SERIES PREDICTION

There are many things that humans find easy to do that computers are currently unable to do. Tasks such as visual pattern manipulating objects by touch, and navigating in a complex world are easy for humans. Yet, despite decades of research, we have no viable algorithms for performing these and other cognitive functions on a computer. In this study, we used a bio-inspired neural network called a KAset neural network to perform a time series predictive task. The results from our experiments showed that the predictive accuracy with this method was better in most markets than results obtained using a random walk method.