New implementation of analog artificial neural network for breast cancer classification

This paper presents a new implementation in FDSOI (Fully Depleted Silicon On Insulator) technology of an Analog ANN (Artificial Neural Network) for breast cancer classification. The architecture of the circuit is based on the MLP (Multi-Layer Perceptron) algorithm with back-propagation. The analog implementation of such an algorithm typically uses multipliers which are surface and power consuming. The second drawback of this topology concerns the storage of the weights. To overcome these problems, we propose to use new current mirrors functioning as multipliers and the advantages of the FDSOI technology. Then, the building blocks of a neuron are described: the new multiplier and the activation function (logsigmoid and its derivative). The final simulation of the ANN exhibits a 2.4% total error in the confusion matrix.