Explainable Machine Learning Models for Healthcare AI


#1

Title: Explainable Machine Learning Models for Healthcare AI
Date: Wednesday, September 26, 2018
Time: 12:00 PM Eastern Daylight Time
Duration: 1 hour

SPEAKER:

Resources:
Webinar Registration Link (free)
AI and Machine Learning for Healthcare (Free Video for ACM Members)
Deep Learning for Health Tech (Free Video for ACM Members)
Learning Path: Machine Learning for Healthcare Using Python, TensorFlow, and R (Free Learning Path for ACM Members)
Big Data Management and the Internet of Things for Improved Health Systems (Free Book for ACM Members)
Healthcare Analytics Made Simple (Free Book for ACM Members)
Ubiquitous Machine Learning and Its Applications (Free Book for ACM Members)
An Introduction to Machine Learning Interpretability (Free Book for ACM Members)
Cybersecurity for Hospitals and Healthcare Facilities: A Guide to Detection and Prevention (Free Book for ACM Members)


#2

Hi, as suggested, I’d like to leave a comment/question for today’s Webinar in advance. Here it is:
As far as I have noticed, IBM and its Watson crews have recently also empasized (or started to emphasize?) “Trust and Transparency for AI” [1]. IBM/Watson proposes or offers (or aims to offer?) the following plan:

  1. Automatically detect bias and analyze its source
  2. (Offer tools to) suggest mitigation strategies
  3. Offer Efficient end-to-end lineage services (whatever that means exactly)
    See even more at [2].

My question (or kind request) to the Speakers of this Webinar would be as follows:

  1. Could the speakers of this Webinar please comment on how they percieve IBM’s state and approach in that regard?
  2. Also, of course, it might be interesting to learn more about the differences between KenSci’s and Watson’s approaches towards “explainability in ML models” (esp. for Healthcare AI).

Thanks a lot in advance! Looking forward to the Webinar. :slight_smile:

[1] https://newsroom.ibm.com/IBM-watson?item=30657
[2] http://www.research.ibm.com/artificial-intelligence/trusted-ai/