AHP, ANP, AND ANN: TECHNICAL DIFFERENCES, CONCEPTUAL CONNECTIONS, HYBRID MODELS

Umberto Gori, Sergio Bedessi, Serena Lisi
2011 unpublished
The AHP (Analytic Hierarchy Process) and the ANP (Analytic Network Process) are formalised decision making methods that examine quantitative and qualitative factors, while ANN (Artificial Neural Networks) can be considered to be data processing systems that are based on how the human brain works, and operate by assigning weights as does a universal function approximator. AHP and ANP are based on an "objective-criteria-alternative" structure; AHP operates hierarchically, while the evolution of
more » ... is model, ANP, introduces the concepts of feedback and interdependence. In this sense, ANP may be seen as a type of "conceptual bridge" between AHP and ANN (when taking into consideration those neural networks that learn through the so-called backpropagation algorithm). This study presents an AHP-ANN hybrid model in which AHP is used during the decision making phase and one or more ANNs substitute the classic pair-wise comparisons typical of AHP assigning weights to the alternatives with regard to one or more of the criteria. Moreover, the study attempts to define the parameters and the case records for which the model is applicable, so as to propose its generalised use rather than ad hoc use for specific cases and situations.
doi:10.13033/isahp.y2011.148 fatcat:5rcn4mfokrhdbbhvm6sh6csbta