Fast Deep Learning Classification With Multilinear Feature Maps
Michel Andre L .Vinagreiro, Edson Caoru Kitani, Armando Antônio Maria Laganá, Leopoldo Rideki Yoshioka · 2022 International Joint Conference on Neural Networks (IJCNN) · 2022
In the late 1990s, [1] and [2] the development and application of Convolutional Neural Networks (CNN). CNN is considered a Deep Learning algorithm and achieved the best performance in image recognition, localization, and segmentation tasks, compared with traditional image processing techniques [3] and [4], mainly due to the CNN ability to extract a large number of features from input images. When [5] won the Imagenet-2012 challenge, a breakthrough occurred, achieving a significant performance improvement over previous architecture. Another successful architecture is the deep neural network proposed in [6], called VGG-16 (Visual Geometry Group), which showed the importance of depth architecture to achieve high-performance classification tasks.