Simon Haykin McMaster University
Much of the signal-processing literature in the 20th Century focused on certain simplifying assumptions: Stationarity, Gaussianity, and linearity. The primary motivation for these assumptions is mathematical tractability. However, with the ever-increasing power and availability of computers, mathematical tractability is no longer the compelling issue that it used to be. Indeed, with powerful computers at hand, we can expand on the horizons of statistical signal processing to account for practical realities of nearly all physical phenomena encountered in practice, namely, nonstationarity, unknown statistics (typically nonGaussian), and nonlinearity. This expansion is one of the major goals of the 2003 Workshop on "New Directions in Signal Processing for the 21st Century".
Another major goal of this unique Workshop is to bring two scientific communities closer together:
The signal processing community, which is still dominated by classical approaches to the solution of real-life problems; and,
The neural computation community, which distinguishes itself by being motivated by the computing power of the human brain but lacks the attention that signal processing problems duly deserve.
With these two goals in mind, and by focusing on a carefully selected set of topics, the Workshop is assured of making a significant difference.
