Multi-Loss Function Fusion for Face Recognition Based on the Convolutional Neural Network
| Autor: | Qihui Peng, Qing Gao, Shibin Xuan, Kaicheng Xiong |
|---|---|
| Rok vydání: | 2019 |
| Předmět: |
Fusion
Computer science business.industry Deep learning 020206 networking & telecommunications Pattern recognition 02 engineering and technology 010501 environmental sciences 01 natural sciences Facial recognition system Convolutional neural network Robustness (computer science) 0202 electrical engineering electronic engineering information engineering Artificial intelligence business 0105 earth and related environmental sciences |
| Zdroj: | CISP-BMEI |
| DOI: | 10.1109/cisp-bmei48845.2019.8965839 |
| Popis: | In recent years, deep convolutional neural network techniques have been widely used in face recognition. Loss function is an important part of convolutional neural network for face recognition. At present, a variety of loss functions have been proposed. Each loss function has its own scope of application as well as advantages and disadvantages. In order to improve the robustness,a face recognition algorithm based on multi-loss function self-adaptive weighted fusion convolutional neural network is proposed. Its core is to learn the weight of each loss function automatically. Experimental results show the improved multi-loss function self-adaptive weighted fusion convolutional neural network can improve the recognition rate of the algorithm effectively. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|