Three accomplished keynote speakers will introduce their research outcomes and intensive experiences.
Keynote #1: Abstraction or Instantiation: On Building the Internet for Smarter Place
(09:20~10:20 Wednesday Oct. 17)
Speaker: Wei-Chao Chen (Skywatch)
There are two ends of the spectrum when it comes to Internet-of-Things. On the one end, the focus is on lower level communication technology, platforms, and standards that aim to enable machine-to-machine communications. On the other end, people have been searching for the killer devices that would transform our lives. Clearly, one is the "Internet" perspective, while the other is the "Things" perspective. In this talk, instead of taking the words "Internet of Things" literally, we focus on the combined effects of devices on environments. Specifically, we discuss the conversations between devices, and how these interactions can open new frontiers of digitized spaces in the era of machine learning.
Keynote #2: Delivering Machine Learning Engineering in Scientific Research
(09:00~10:00 Thursday Oct. 18)
Speaker: David Chiu (LargitData)
As the use of machine learning to train a predictive model has become more and more popular, knowing how to build a model with a fixed dataset may not be news to most researchers nowadays. With a little programming training, any researcher can now easily create a model with just a few lines of coding with some state of the art programming languages and analysis tool. However, knowing how to build the model is not enough, if you wish your research can reach to a wider audience, it would be important to know the correct approach to deliver the learned model into a production environment. In this keynote speech, I will address how to bring machine learning model into the production environment from the perspective of a data scientist and engineer. I will cover the issues includes data preparation, model versioning, model deployment, continuous integration/delivery, model validation, how to choose evaluation metrics, some challenges when facing big data.
Keynote #3: Bootstrapping AI Adoption for Smart Manufacturing
(09:00~10:00 Friday Oct. 19)
Speaker: Trista Chen (Inventec)
When it comes to applying AI to smart manufacturing, the first problems that come to mind are usually the automatic inspection of products. The reality is we often need to carefully re-define the problems to yield a satisfactory solution and return on investment. In this talk, we share our experiences working with first-tier electronics manufacturing facilities to apply machine learning techniques to their product design process and production pipeline. We will also talk in more detail about several applications in the domains of circuit defect and product appearance inspections, quality control, and the associated edge AI algorithms and machinery we have created to solve these problems.
