殘差神經網絡
閱讀設定
殘差神經網絡(residual neural network,ResNet)係指用咗「跳過連繫」(skip connection)嘅人工神經網絡。「跳過連繫」意思即係話有部份嘅人工神經細胞可以唔淨只連去下一層嘅細胞嗰度,仲可以「跳過」若干層連去(例如)兩層後或者三層後嘅細胞。殘差神經網絡係基於生物神經網絡嘅-現實嘅生物神經網絡好少可會好似廿世紀嘅前饋神經網絡噉完美清楚分層[1]。
攷
[編輯]- ↑ He, Kaiming; Zhang, Xiangyu; Ren, Shaoqing; Sun, Jian (2016). "Deep Residual Learning for Image Recognition" (PDF). Proc. Computer Vision and Pattern Recognition (CVPR), IEEE.