《Large-Margin Softmax Loss for Convolutional Neural Networks》
2017-01-20 04:44
369 查看
机器学习中的一个任务是学习具有判别性的特征,使得类内距离较小,类间距离较大,或者说使得学习到的特征具有类内的紧致性和类间的分离性。传统的做法是利用contrastive loss,triplet loss,还有最近提出来的center loss+softmax。参考论文提出了large-margin softmax (L-Softmax) loss也是以此为目的,它扩展了传统的Softmax,取得了更好的效果。
本文持续更新!如文中有错误,或你对本文有疑问或建议,欢迎留言!
2017年01月20日,发表博文。
更新记录
康凯@Dilusense本文持续更新!如文中有错误,或你对本文有疑问或建议,欢迎留言!
2017年01月20日,发表博文。
参考文献
[2016 ICML] Large-Margin Softmax Loss for Convolutional Neural Networks相关代码
import numpy as np import matplotlib.pyplot as plt def angleMargin(m): assert m > 0 and isinstance(m, int) pts = 100 psi = np.empty(m*pts) theta = np.empty(m*pts) for k in range(m): t = np.linspace(k*np.pi/m, (k+1)*np.pi/m, pts) theta[k*pts:(k+1)*pts] = t psi[k*pts:(k+1)*pts] = (-1)**k*np.cos(m*t) - 2*k return theta, psi if __name__ == '__main__': for m in range(1, 5): theta, psi = angleMargin(m) plt.plot(theta, psi, label = 'm = {}'.format(m)) plt.legend(loc = 'best') plt.xlabel(r'$\theta$') plt.ylabel(r'$\psi(\theta)$')
正文
相关文章推荐
- Large-Margin Softmax Loss for Convolutional Neural Networks
- 大裕量Softmax 损失函数--Large-Margin Softmax Loss for Convolutional Neural Networks
- 阅读小结:Large-Margin Softmax Loss for Convolutional Neural Networks
- 【论文笔记】Large-Margin Softmax Loss for Convolutional Neural Networks
- 论文阅读:ICML 2016 Large-Margin Softmax Loss for Convolutional Neural Networks
- 基于Caffe的Large Margin Softmax Loss的实现(中)
- 基于Caffe的Large Margin Softmax Loss的实现(上)
- Sampled Softmax 论文笔记:On Using Very Large Target Vocabulary for Neural Machine Translation
- Sampled Softmax 论文笔记:On Using Very Large Target Vocabulary for Neural Machine Translation
- 损失函数改进之Large-Margin Softmax Loss
- A guide to receptive field arithmetic for Convolutional Neural Networks
- [译]CS231n 卷积神经网络对于图像识别的应用--(一)(CS231n Convolutional Neural Networks for Visual Recognition)
- 【论文学习】Large-scale Video Classification with Convolutional Neural Networks
- Very Deep Convolutional Networks for Large-Scale Image Recognition
- UNDERSTANDING CONVOLUTIONAL NEURAL NETWORKS FOR NLP
- deeplearning论文学习笔记(1)Convolutional Neural Networks for Sentence Classification
- 深度学习研究理解:Very Deep Convolutional Networks for Large-Scale Image Recognition
- Spatially Supervised Recurrent Convolutional Neural Networks for Visual Object Tracking
- Convolutional Neural Networks for Speech
- CNN网络量化 - Quantized Convolutional Neural Networks for Mobile Devices