谷歌推出了AdaNet，用于将算法与AutoML相结合
谷歌今天推出了AdaNet，这是一种开源工具，用于结合机器学习算法，以获得更好的预测见解。AdaNet 今天在Tensor GitHub存储库中可用。
谷歌人工智能软件工程师查尔斯威尔在一篇博客文章中说：“AdaNet建立在我们最近的强化学习和基于进化的AutoML努力的基础上，在提供学习保证的同时快速灵活。”
“重要的是，AdaNet提供了一个通用框架，不仅可以学习神经网络架构，还可以学习集成以获得更好的模型。”
AdaNet使用一种称为集成学习的方法来组合和改进算法，这种方法以前需要领域专业知识或者需要太多时间进行培训，Weill说。
为了更容易实现AdaNet，该框架插入TensorFlow Estimator以将基本信息集中到一个地方，以及 TensorBoard，它在训练AI模型时提供视觉反馈。
AdaNet通过学习神经网络的架构，然后为其添加子网，确保其创建的集合模型的学习保证。
想要更多控制流程的机器学习从业者可以使用TensorFlow API定义自己的子网，自定义丢失功能或切换其他设置。
关于AdaNet如何运作的其他细节可以在去年在机器学习国际会议上发表的这篇发表的论文中看到。
今天发布的AdaNet是AutoML的最新进展，这是Google自动培训和部署神经网络的方式。Google
Cloud Platform在今年夏天推出了用于翻译，计算机视觉和自然语言处理的 AutoML，以及用于在1月份构建自定义AI模型的
Cloud AutoML 。
{"weixin":{"label":"微信","name":"weixin","selected":true,"value":true,"sortid":"1","shareid":"weixin","sharetitle":"分享到微信","event":"shareToWeiXin","lang":"shareWeb_WeiXin"},"copy":{"label":"复制网址","name":"copy","selected":true,"value":true,"sortid":"2","shareid":"copy","sharetitle":"复制网址","event":"copy_url","lang":"shareWeb_Copy"},"qq":{"label":"QQ好友","name":"qq","selected":true,"value":false,"sortid":"1","shareid":"qq","sharetitle":"分享到QQ","event":"shareToQQ","lang":"shareWeb_QQ"},"sina_weibo":{"label":"新浪微博","name":"sina_weibo","selected":true,"value":true,"sortid":"4","shareid":"sina_weibo","sharetitle":"分享到新浪微博","event":"shareToSinaWB","lang":"shareWeb_SinaWeiBo"},"qq_zone":{"label":"QQ空间","name":"qq_zone","selected":true,"value":true,"sortid":"5","shareid":"qq_zone","sharetitle":"分享到QQ空间","event":"shareToQzone","lang":"shareWeb_QQZone"},"renren":{"label":"人人网","name":"renren","selected":true,"value":true,"sortid":"7","shareid":"renren","sharetitle":"分享到人人网","event":"shareToRenren","lang":"shareWeb_RenRen"},"douban":{"label":"豆瓣网","name":"douban","selected":true,"value":true,"sortid":"8","shareid":"douban","sharetitle":"分享到豆瓣网","event":"shareToDouban","lang":"shareWeb_DouBan"},"baidu_tieba":{"label":"百度贴吧","name":"baidu_tieba","selected":true,"value":true,"sortid":"10","shareid":"baidu_tieba","sharetitle":"分享到百度贴吧","event":"shareToTieba","lang":"shareWeb_TieBa"},"Facebook":{"label":"Facebook","name":"Facebook","selected":true,"value":true,"sortid":"11","shareid":"Facebook","sharetitle":"分享到FaceBook","event":"shareToFacebook","lang":"shareWeb_Facebook"},"Twitter":{"label":"Twitter","name":"Twitter","selected":true,"value":true,"sortid":"12","shareid":"Twitter","sharetitle":"分享到Twitter","event":"shareToTwitter","lang":"shareWeb_Twitter"},"LinkedIn":{"label":"LinkedIn","name":"LinkedIn","selected":true,"value":true,"sortid":"13","shareid":"LinkedIn","sharetitle":"分享到linkedIn","event":"shareToLinkedin","lang":"shareWeb_Linkedin"},"whatsapp":{"label":"whatsapp","name":"whatsapp","selected":true,"value":true,"sortid":"15","shareid":"whatsapp","sharetitle":"分享到whatsapp","event":"shareToWhatsapp","lang":"shareWeb_whatsapp"},"line":{"label":"line","name":"line","selected":true,"value":true,"sortid":"15","shareid":"line","sharetitle":"分享到line","event":"shareToLine","lang":"shareWeb_line"},"qq_weibo":{"label":"腾讯微博","name":"qq_weibo","selected":true,"value":true,"sortid":"3","shareid":"qq_weibo","sharetitle":"分享到腾讯微博","event":"shareToQQwb","lang":"shareWeb_QQWeiBo"},"peopleBlog":{"label":"人民微博","name":"propleBlog","selected":true,"value":true,"sortid":"14","shareid":"propleBlog","sharetitle":"分享到人民微博","event":"shareToPeopleBlog","lang":"shareWeb_peopleBlog"}}
