OneClick.aiCEO

个人简介：Yuan Shen is the founder and CEO of OneClick.ai, the first Automatic Deep Learning platform that automatically builds, trains, and deploys AI models. Yuan has more than a decade of experience in Machine Learning across medical imaging, e-commerce, online advertising, and search engines. Prior to OneClick.ai, Yuan worked at Microsoft and eBay.

 

议题：自动化深度学习开创人工智能的未来Automatic Deep Learning Defines The Future of AI

 

议题简介：Today’s business leaders are faced with a dilemma about AI. No one wants to miss the opportunity to stay ahead of the game, but implementing AI is a huge investment that comes with a lot of risks. According to Gartner, 60% of AI projects failed in US businesses in 2017. Not to mention the increasing gap in AI expertise. This is an even greater challenge when it comes to deep learning, the latest technology that has brought AI back in the spotlight. The benefits of using deep learning are enormous, as it works much better with unstructured data (images, text, or mix of them combined with numerical ones) than any technology predating it. However, few data scientists in the market have the skills to build the customized deep neural networks needed to unleash its value. Automated Deep Learning (AutoDL) is giving hope to easier and risk-free deep learning applications. AutoDL relies on advanced algorithms to design and deploy deep neural networks automatically, allowing rapid development of deep learning applications in days instead of months or years. It doesn’t require technical expertise from its users and enables innovators to iterate through ideas in days to realize business impact with minimum effort. OneClick.ai CEO Yuan Shen will share technical insights into various AutoDL technologies and implementations. He will also highlight business applications enabled by AutoDL like omni-channel sales forecasting, product image recognition, and dynamic pricing. The session is designed to leave you with a clearer picture about the foreseeable future of AI and ideas for how your organization can take advantage of the latest AutoDL advancements—not in the future, but today.

 

11日培训题目：深度学习时间序列分析和实战

 

培训大纲：时间序列预测做为一种重要的预测手段，在各个领域都有重要应用。对于零售产业来说，一方面需要销售预测来提前制定销售计划；另一方面也需要准确的销售预测来为供应链管理提供指导。对航空旅游行业来说，基于实时的需求预测也是企业最大化利润的关键。在工业界，基于时序预测的智能运维也在逐渐提高生成环节的稳定性。传统时序预测常常采用ARIMA或者指数平滑等线性模型，对于日渐复杂的真实生成运营环境常常不能满足需求。深度学习在图像处理，自然语言理解和语音识别等领域取得了突破性进展，已经成为人工智能的一个热点。深度学习由于强大的非线性特征表征能力使得其与时间序列预测结果成为新的热点。

 

9:00-10:30am: 传统时序模型：AR，MA，SARIMA以及他们的问题和挑战。随着现在各预测的问题复杂化，传统模型往往很难处理非常复杂的关系。我们同时会聊到如何用传统的机器学习算法来处理时序预测。

10:30-11:30am: 深度学习中处理序列的模型：RNN，LSTM，以及WaveNet。我们也会分析如何用深度学习来处理复杂的时序预测问题。

11:30-12:00pm: 讲解自动化的机器学习和深度学习技术（AutoML和AutoDL），以及他们如何能够提高建立时间序列模型的效率，更方便的完成人工智能项目的开发和部署。

 

适合人群：机器学习和深度学习技术人员，对自动化机器学习感兴趣技术开发人员。

 

难度：入门到中等。

 

提前掌握知识点：机器学习知识基础