Semi-Tensor-Product Based Convolutional Neural Networks
This work addresses a specific issue in CNN design for researchers in signal processing and computer vision, but it appears incremental as it builds on existing semi-tensor product concepts.
The paper tackled the problem of junk information in convolutional neural networks caused by padding by proposing a new convolutional product based on the semi-tensor product, which avoids padding. It applied this method to image and third-order signal identification tasks, though no concrete performance numbers were provided.
The semi-tensor product (STP) of vectors is a generalization of conventional inner product of vectors, which allows the factor vectors to of different dimensions. This paper proposes a domain-based convolutional product (CP). Combining domain-based CP with STP of vectors, a new CP is proposed. Since there is no zero or any other padding, it can avoid the junk information caused by padding. Using it, the STP-based convolutional neural network (CNN) is developed. Its application to image and third order signal identifications is considered.