| 标题 |
[高分] 学位论文 Optimizations on ResNet Using a Self-Attention Mechanism
基于自我注意机制的ResNet优化
|
| 网址 |
求助人暂未提供
|
| DOI |
暂未提供，该求助的时间将会延长，查看原因？
|
| 其它 |
摘要 The go-to method applied to computer vision tasks involves convolutional neural networks. While computer vision has made great strides in utilizing convolutional neural networks there is still room for optimization. Many researchers still attempt to improve accuracy as well as improve computational requirements of the networks. ResNet is one such model which set out to improve the capabilities of these networks. In this work, an extension of the residual-unit from ResNet is introduced utilizing a self-attention mechanism. Using this mechanism, we have attempted to improve the classification accuracy of ResNet on the CIFAR-100 dataset, as well as to use this mechanism as a tool to determine the most significant portions of a ResNet. Attempts to improve the performance of ResNet was inconclusive on CIFAR-100, but using this mechanism to determine effective regions of the network to prune has been successful. |
| 求助人 | |
| 下载 | 求助已完成，仅限求助人下载。 |
毛豆
无名老大
归尘
方羽
36456657
pluto
Leonardi
南宫清涟
坚强亦丝
彭于彦祖
iNk
Candice
pcr163
杳鸢
hardworkcd
跳跃仙人掌
Singularity
尊敬的金针菇
tuanheqi
快乐滑板
皖公网安备34019202002308
