没有合适的资源?快使用搜索试试~ 我知道了~
第二周写作作业A Proposal to Implement optimized reward functions NetAdapt to searching
资源详情
资源评论
资源推荐
第二周写作作业
A Proposal to Implement optimized reward functions NetAdapt to searching for
a light-weight backbone for tow-stage detectors.
Summary
Statement of Problem
Network architecture and hyper-parameter optimization is a black-box
optimization problem, and the recently popular search strategy defines what
algorithms can be used to quickly and accurately find the optimal configuration
of network structure parameters. A practical problem is that searching for a
optimized structure is time consuming and computational expensive. We are
expected to optimize the reward functions in NetAdapt and propose a more
efficient backbone for two-stage detectors.
Related work
The previous methods of model structure selection
The related research on the trade off between low latency and accuracy
The development of neural network architecture search
Objectives
An optimized reward functions for NetAdapt
A new two-stage detector with Feature Pyramid and a light-weight backbone
Plan of Action
Indentifying the detector framework for the research
Evaluate the iteration time using NetAdapt
Developing appropriate reward functions
Selecting the network structure generated within a reasonable time
三更寒天
- 粉丝: 79
- 资源: 326
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功
评论0