• Transportation Research Board

      TRB 98th Annual Meeting

      January 13–17, 2019

    • Interactive Program


Annual Meeting Event Detail

Workshop 1058

Big Data Without Machine Learning Is Just Lots of Data: A Guided Tour to Big Data and Machine Learning

Sunday 1:30 PM- 4:30 PM
Sign in to reveal location
David Reinke, Kittelson & Associates, Inc. (KAI), presiding
Mecit Cetin, Old Dominion University, presiding
Sponsored by:
Standing Committee on Artificial Intelligence and Advanced Computing Applications (ABJ70)

Past workshops on big data have focused on data storage and traditional statistics. This workshop emphasizes the connection between big data and machine learning, providing an overview of machine learning essentials, mathematical and statistical background, and applications to transportation including automated vehicles. Methods will include supervised and unsupervised learning as well as deep learning. Also presented will be an overview of the Artificial Intelligence and Advanced Computing Applications Committee’s forthcoming primer on machine learning.

1         Introduction and overview

Time: 15 min

Presenter: David Reinke

  • What is machine learning
  • Where machine learning fits into AI
  • Types of machine learning: supervised, unsupervised, reinforcement
  • Supervised methods: classification, regression
  • Importance of unsupervised learning: preponderance of unlabeled data
  • Machine learning vs. traditional statistics
  • Why machine learning is necessary for big data
  • Who this workshop is for: transportation students, academics, practitioners
  • ABJ70 machine learning primer
  • Overview of rest of workshop

2         A brief tour of machine learning methods

Time: 30 min

Presenter: TBD

Potential topics. Just give brief description of what they are, what they can do, advantages and disadvantages. Should cover at least the ones most often used and cutting-edge methods (NN, SVM, deep learning, clustering) plus perhaps 2 – 3 (at most) others.

  • Neural networks
  • Support vector machines
  • Deep learning
  • Unsupervised methods / clustering
  • Other methods (2 – 3 at most): e.g., boosting, relevance vector machines

3         Tools of the trade: math, stat, and all that

Time: 20 min

Presenter: David Reinke

Brief overview of required background and tools for machine learning. Focus on concepts rather than details.

  • Background knowledge
    • Mathematics
    • Information theory
    • Numerical analysis
    • Statistics (Bayesian)
  • Software for machine learning
    • Programming languages. Procedural vs OOP vs functional
    • Software packages for machine learning (R, Python, TensorFlow, Julia, etc.)

Break – 10 min

4         Applications of machine learning – results from the TRANSFOR19 data competition sponsored by Didi Chuxing

Time: 70 min

Presenters: TBD

Three presenters, each discussing a particular application of machine learning: data sources, methods used, results. Should also emphasize advantage of machine learning approach over traditional statistical methods.

5         Forthcoming outreach efforts from ABJ70

Time: 5 min

Presenter: TBD

  • Resources guide
  • Machine learning primer
  • Webinars

6         Discussion / Q&A

Time: 30 min

Title Presentation Number
Introduction and Overview
David Reinke, Kittelson & Associates, Inc. (KAI)
Tools of the Trade: Mathematics, Statistics, and All That
David Reinke, Kittelson & Associates, Inc. (KAI)