An autotuning method for the optimum sigmoid function of neural networks is proposed. It is based on the steepest descent method. Simulated results using a learning-type direct controller confirm both the practicality and the characteristics of the autotuning method.
A method is proposed for preventing privacy invasion through unintentional capture of facial images. Prevention methods such as covering the face and painting particular patterns on the face are effective but hinder face-to-face communication. The proposed method overcomes this problem through the use of a device worn on the face that transmits… (More)