Integrating Neural Networks and Rule Based Systems to build anAvalanche Forecasting System.
MARTIN P. SCHWEIZER, Philipp Foehn, Juerg Schweizer · 1994
A new approach to the problem of avalanche forecasting is presented. A hybrid expert system called ALUDES 1 was developed to assess the avalanche danger for a given region. Using snow, weather and snow cover data as input parame ters the system evaluates the degree of danger for a given region. It integrates extended symbolic computing from tra ditional Artificial Intelligence and connectionist methods using Kohonen Networks. By generating (symbolic) fuzzy rules with the fitge 2 -method from subsymbolic data through