It's now time to conclude our discussion on automation with a short introduction to Automated ML or AutoML in the final section. Summary In the past decade, deep learning has made incredible progress. Every as Auto Keras, Auto WEKA, NNI and auto-sklearn have gone even further and adopted the doctrine of AutoML which aims to automate most of the steps involved in the machine learning pipelines to reduce the the dependency on ML experts and to promote large-scale adoption of machine learning. An AutoML pipeline assumes all the responsibilities which traditionally required ML experts. Imagine that we are developing

Showcase AI SaaS Showcase AI 通用物品识别平台架构 品 识 AI 中 台 AI 算法库 AI 核心模块 AI 行业模型 数据集 模型训练 模型管理 AutoML AI 物品库 服务管理 模型压缩 棚格图识别 货架巡检 商品推荐 陈列审核 入库审计 货物盘点 构件识别 CAD解析 规则审查 户型图识别 视频盘点 自动分拣 细粒度识别

128,64,10]，可以自由配置，如[256,256,64,10]或 [512,64,32,10]等都是可行的。至于哪一组超参数是最优的，这需要丰富的领域经验知识和 大量的实验尝试，或者可以通过 AutoML(Auto Machine Learning)技术搜索出较优设定。 输入：[?, 784] 隐藏层1：[256] 隐藏层2：[128] 隐藏层3：[64] 输出层：[?, 10] 图 6