Covolutional Multinomial Logistic Regression for Face Recognition
Surachai Ongkittikul, Jirawut Suwatcharakulthorn, Kanoksom Chutisowan, Kongnat Ratanarangsank · 2020
the system of machine learning in term of supervised learning at present. It is popular for building learning systems for recognition. Or classify objects within the image. This paper therefore presents methods and systems for enhancing efficiency in order to select face recognition with multinomial logistic regression (MLR). Which will use methods of filtering with information before bringing it into the learning process. Which starts by Gaussian Filtering that convolute with the image for smoothing image. Then reduce the pixel's amount and manage data that causes max pooling to be chosen in this paper. Next, it was used to create a model to find the weight of the face image data. And last, test the performance of the work to see the error test with weight that was calculated from MRL. The experimental in the paper was tested the creation of data on the face database AT&T, which is the most widely used face database.