Hyperellipsoidal neuron

摘要: In recent years, the research on neural networks has been guided by search of new mathematical frameworks, with hope finding features, as geometric interpretation, for facing today problems or reducing computational cost. this paper we introduce a Clifford Neuron [1], extending conformai neuron, presented in [2] through generalization algebra quadratic surfaces (G 6,3 ), [3]. can obtain decision different shapes, depending input data: spherical surface, ellipsoidal, cylindrical even surface pair planes (all them be derived special case an ellipse). The above without need using kernel technique, just linear activation function over hiperconformal space.