First European Conference on Speech Communication and Technology

Paris, France
September 27-29, 1989

New Backpropagation Algorithm Using Quadratic Potential Functions, and an Experiment on Isolated Word Recognition

Enric Monte, Eduardo Lleida, Jose B. Marino

Dpto. Teoria de la Senal y Comunicaciones, Universidad Politecnica de Cataluna, Spain

This paper presents a new algorithm to train multilayered perceptrons, using quadratical potential functions. This new algorithm is compared in an isolated word recognition task, with the back propagation algorithm that uses linear combinations of the inputs. Some pattern recognition techniques are also used to reduce the dimensinality of the input pattern in order to reduce the computational burden of the training and the recognition. The algorithm that uses quadratic potential functions yields better results in the recognition task.

Full Paper

Bibliographic reference.  Monte, Enric / Lleida, Eduardo / Marino, Jose B. (1989): "New backpropagation algorithm using quadratic potential functions, and an experiment on isolated word recognition", In EUROSPEECH-1989, 1341-1344.