深度学习论文中常用的词汇

小咪咪 2021-12-13 13:39 424阅读 0赞

卷积神经网络中的常用词：

<table style="width:500px;"> 
 <tbody> 
  <tr> 
   <td style="width:227px;">feed-forward</td> 
   <td style="width:271px;">前馈运算</td> 
  </tr> 
  <tr> 
   <td style="width:227px;">objective function</td> 
   <td style="width:271px;">目标函数</td> 
  </tr> 
  <tr> 
   <td style="width:227px;">back propagation</td> 
   <td style="width:271px;">反向传播</td> 
  </tr> 
  <tr> 
   <td style="width:227px;">back-forward</td> 
   <td style="width:271px;">反馈</td> 
  </tr> 
  <tr> 
   <td style="width:227px;">deep belief network</td> 
   <td style="width:271px;">深度置信网络</td> 
  </tr> 
  <tr> 
   <td style="width:227px;">recurrent neural network</td> 
   <td style="width:271px;">递归神经网络</td> 
  </tr> 
  <tr> 
   <td style="width:227px;">overfitting</td> 
   <td style="width:271px;">过拟合</td> 
  </tr> 
  <tr> 
   <td style="width:227px;">activation function</td> 
   <td style="width:271px;">激活函数</td> 
  </tr> 
  <tr> 
   <td style="width:227px;">ground truth</td> 
   <td style="width:271px;">真实标记</td> 
  </tr> 
  <tr> 
   <td style="width:227px;">&nbsp;</td> 
   <td style="width:271px;">&nbsp;</td> 
  </tr> 
  <tr> 
   <td style="width:227px;">&nbsp;</td> 
   <td style="width:271px;">&nbsp;</td> 
  </tr> 
  <tr> 
   <td style="width:227px;">&nbsp;</td> 
   <td style="width:271px;">&nbsp;</td> 
  </tr> 
 </tbody> 
</table>

《[Scale-Aware Trident Networks for Object Detection][]》

——关于这篇论文的说明：[TridentNet：处理目标检测中尺度变化新思路][TridentNet]

<table style="width:500px;"> 
 <tbody> 
  <tr> 
   <td style="width:229px;">multi-scale</td> 
   <td style="width:269px;"> <p>多尺度</p> </td> 
  </tr> 
  <tr> 
   <td style="width:229px;">pyramid</td> 
   <td style="width:269px;">金字塔</td> 
  </tr> 
  <tr> 
   <td style="width:229px;">down sample rate</td> 
   <td style="width:269px;">降采样率</td> 
  </tr> 
  <tr> 
   <td style="width:229px;">receptive field</td> 
   <td style="width:269px;">接收场</td> 
  </tr> 
  <tr> 
   <td style="width:229px;"><span style="color:#f33b45;"><strong>receptive field</strong></span></td> 
   <td style="width:269px;"><span style="color:#f33b45;">感受野;</span></td> 
  </tr> 
  <tr> 
   <td style="width:229px;">backbone</td> 
   <td style="width:269px;">骨干，基础</td> 
  </tr> 
  <tr> 
   <td style="width:229px;">dilation rate</td> 
   <td style="width:269px;">膨胀率</td> 
  </tr> 
  <tr> 
   <td style="width:229px;">trident block</td> 
   <td style="width:269px;">三叉戟块</td> 
  </tr> 
  <tr> 
   <td style="width:229px;">parallel multi-branch</td> 
   <td style="width:269px;">并行多分支</td> 
  </tr> 
  <tr> 
   <td style="width:229px;">ablation analyses</td> 
   <td style="width:269px;">消融分析</td> 
  </tr> 
  <tr> 
   <td style="width:229px;">&nbsp;</td> 
   <td style="width:269px;">&nbsp;</td> 
  </tr> 
  <tr> 
   <td style="width:229px;">&nbsp;</td> 
   <td style="width:269px;">&nbsp;</td> 
  </tr> 
 </tbody> 
</table>

以resnet为基础的trident模型源码：[https://github.com/TuSimple/simpledet/tree/master/models/tridentnet][https_github.com_TuSimple_simpledet_tree_master_models_tridentnet]

[Scale-Aware Trident Networks for Object Detection]: https://link.zhihu.com/?target=https%3A//arxiv.org/abs/1901.01892
[TridentNet]: https://zhuanlan.zhihu.com/p/54334986
[https_github.com_TuSimple_simpledet_tree_master_models_tridentnet]: https://github.com/TuSimple/simpledet/tree/master/models/tridentnet