基于深度学习的时间序列预测 (Deep learning for time series forecasting)

中英文讲话(Presented in Chinese and English)

Yijing Chen (Microsoft), Dmitry Pechyoni (Microsoft), Angus Taylor (Microsoft), Vanja Paunic (Microsoft), Henry Zeng (Microsoft)
09:0012:30 Wednesday, June 19, 2019
中英文讲话(Presented in Chinese and English)
Location: 报告厅(Auditorium)
Average rating: *****
(5.00, 2 ratings)

必要预备知识 (Prerequisite Knowledge)

  • Familiarity with machine learning and neural networks
  • Experience building machine learning models

您将学到什么 (What you'll learn)

  • Understand the basics of convolutional neural networks (CNNs) and recurrent neural networks (RNNs) and their advanced architectures that are effective in time series forecasting
  • Learn when to use deep learning models instead of traditional time series models in time series forecasting
  • Explore a number of techniques and tricks that are important for building successful deep learning models for time series forecasting
  • Get access to the source code of examples of how to train and tune deep learning models for time series forecasting using Keras

描述 (Description)

Time series forecasting is one of the most important topics in data science. Almost every business needs to predict the future in order to make better decisions and allocate resources more effectively. You may need to use time series forecasting for financial forecasting, product sales forecasting, web traffic forecasting, energy demand forecasting for buildings and data centers, or many other reasons. However, most existing forecasting solutions use traditional time series and machine learning models. For your complex forecasting problems, you need to know how to leverage advanced techniques to generate more accurate forecasts. Deep neural networks have achieved a lot of success for many applications. In particular, RNNs are frequently used in text, speech, and video analysis, being designed for processing sequential data. Additionally, CNNs have achieved state-of-the-art performance on many computer vision tasks. These methods have only recently been applied to the task of time series forecasting but typically deliver strong performance.

Yijing Chen, Dmitry Pechyoni, Angus Taylor, Vanja Paunic, and Henry Zeng explore the basic concepts for building such models and show you how and when to apply them to time series forecasting. Once you have a quick overview of time series forecasting and neural networks and a clear background on the problem, Yijing, Dmitry, Angus, Vanja, and Henry dive deep into CNNs and RNNs, including long short-term memory (LSTM) and GRU. At the end of each introduction, you’ll solidify the basics with a hands-on exercise, then move into more advanced topics. You’ll learn how to apply encoder-decoder RNN architecture to time series forecasting, build state-of-the-art hybrid models with CNNs and RNNs, and do hyperparameter tuning for such advanced models in real application.


Tutorial introduction

  • Tutorial goals
  • Tutorial agenda
  • Target audience
  • Why use a neural networks model for time series forecasting?

Knowledge recap

  • Time series recap
    • Time series and time series forecasting
    • Why is time series forecasting important?
    • Questions to ask before building forecast model
  • Feedforward neural network recap
    • Perceptron and multilayer perceptron​
    • Backpropagation​

Introduction to CNNs

  • Introduction to CNNs
  • How CNNs are trained
  • Hands-on exercise: How to apply CNNs to time series forecasting

Introduction to RNNs

  • What are RNNs? ​
  • How RNNs are trained: Backpropagation through time (BPTT)​
  • Vanilla RNN and its gradient problems​
  • Other RNN units​
    • LSTM​
    • GRU​
  • RNN stacking
  • Hands-on exercise: How to apply RNNs to time series forecasting

Encoder-decoder RNN model

  • The encoder-decoder RNN architecture
  • How to apply encode-decode RNN to multistep time series forecasting

Build state-of-the-art models for time series forecasting

  • Hybrid CNN and RNN
  • Hybrid traditional time series and RNN
  • Hybrid RNN and feedforward
  • Hands-on exercise

Hyperparameter tuning

  • How to do hyperparameter tuning
  • Hands-on exercise

Conclusion and key takeaways

Photo of Yijing Chen

Yijing Chen


Yijing Chen is a senior data scientist in the Cloud AI Group at Microsoft, where she works with external customers in areas such as energy demand forecast, user mobile behavioral analysis, retail demand forecast, energy theft detection, product pricing, and medical claim denial prediction as well as on other projects using various machine learning methods. Yijing holds an MA in statistics from Harvard University.

Photo of Dmitry Pechyoni

Dmitry Pechyoni


Dmitry Pechyoni is a senior data scientist in the Cloud AI Group at Microsoft, where he works on building end-to-end data science solutions in various domains, including retail, energy management, and predictive maintenance. Previously, he built machine learning models for display advertising Akamai and MediaMath. Dmitry holds a PhD in theoretical machine learning from the Technion – Israel Institute of Technology.

Photo of Angus Taylor

Angus Taylor


Angus Taylor is a data scientist in the Cloud AI Group at Microsoft, where he builds data science solutions for external customers in the retail, energy, engineering, and package distribution sectors. He holds an MSc in AI from the University of Edinburgh.

Photo of Vanja Paunic

Vanja Paunic


Vanja Paunic is a data scientist in the Algorithms and Data Science Group at Microsoft London. She works on building machine learning solutions with external companies utilizing Microsoft’s AI Cloud Platform. She holds a PhD in computer science with a focus on data mining in the biomedical domain from the University of Minnesota.

Photo of Henry Zeng

Henry Zeng


Henry Zeng is a principal program manager on the AI platform team at Microsoft, where he works with the engineering team, partners, and customers to ensure AzureML is the best ML platform in the cloud. He’s been in the AI and data area for more than 14 years in areas such as database, big data, machine learning, and deep learning. Previously, he was the lead AI solution architect at Microsoft China, where he worked with partners and customers to land AI solutions in manufactory, retail, finance, education, and public service. Henry holds an MS in computer science from Wuhan University.

Comments on this page are now closed.


Picture of Henry Zeng
2019-05-13 23:14 CST

Yes, major parts of the session will be presented in English. Looking forward to meeting you there.

2019-05-11 05:22 CST

Topic looks really good. Is there a portion presented in English?

Picture of Henry Zeng
2019-04-05 02:18 CST

Thanks Yunyi, hope to see you in the tutorial and discuss with you in face over those topics.

2019-04-03 21:38 CST

Dear Speakers,

I am interested in how deep learning can help to forecast time series with a hybrid of multiple seasonalities (e.g. yearly, weekly, daily) and external regressors (e.g. holidays, campaigns, forecasted weather). It would be great if you shed some light on this topic. Thank you!