Ontolog Forum
Session | Explainable AI |
---|---|
Duration | 1 hour |
Date/Time | Apr 10 2019 16:00 GMT |
9:00am PDT/12:00pm EDT | |
5:00pm BST/6:00pm CEST | |
Convener | Todd Schneider |
Ontology Summit 2019 XAI Session 3
Agenda
- Prof. Sargur Srihari
- SUNY Distinguished Professor in the Department of Computer Science and Engineering at the University at Buffalo
- "Explainable Artificial Intelligence: The Probabilistic Approach"
- Bio: https://cedar.buffalo.edu/~srihari/
Conference Call Information
- Date: Wednesday, 10-April-2019
- Start Time: 9:00am PDT / 12:00pm EDT / 6:00pm CEST / 5:00pm BST / 1600 UTC
- ref: World Clock
- Expected Call Duration: 1 hour
- The Video Conference URL is https://zoom.us/j/689971575
- iPhone one-tap :
- US: +16699006833,,689971575# or +16465588665,,689971575#
- Telephone:
- Dial(for higher quality, dial a number based on your current location): US: +1 669 900 6833 or +1 646 558 8665
- Meeting ID: 689 971 575
- International numbers available: https://zoom.us/u/Iuuiouo
- iPhone one-tap :
- Chat Room
Attendees
Proceedings
[12:14] ToddSchneider: Ravi can you act as host. I'm not able to get microphone to work.
[12:17] RaviSharma: Todd yes i will be happy thanks
[12:19] ToddSchneider: Ravi, make sure to ask Sargur to send his slides to Ken (for posting on meeting page).
[12:23] John Sowa: Human-level explanations are always dialogs.
[12:24] John Sowa: A single Q/A is a very rare case. Follow-up questions and discussion are essential.
[12:28] John Sowa: The idea of first, second, and third wave implies that one replaces another.
[12:29] John Sowa: But it's more important to have a tool kit that includes *every* method in a mix & match system.
[12:31] RaviSharma1: John I agree that way one can choose the wave
[12:34] John Sowa: But how can an AI system learn what features are relevant?
[12:34] John Sowa: The most difficult task is to build the model.
[12:35] John Sowa: How can model building be automated?
[12:42] John Sowa: How could any such method be applied to a driverless car?
[12:43] John Sowa: When you're driving down the highway, you may have to make a decision in a split second. How do you derive the model and compute the probabilities?
[12:44] RaviSharma1: Q why log, also reason for Gausian distribution?
[12:44] John Sowa: I was thinking about the recent new about Seattle.
[12:45] John Sowa: one mile of telephone poles fell across a highway.
[12:45] John Sowa: The probability was o.ooooooooooooooooooooooooo1
[12:46] John Sowa: but when it happens, you have to deal with it.
[12:46] RaviSharma1: it would be detectable if size of poles are obstructions and AI should be able to brake the car
[12:48] janet singer: Ravi - yes, in that case the physical obstruction could be handled as such without explanation or understanding
[12:50] Ram D. Sriram: @Hari: What about the scalability of probabilistic network-based systems
[12:52] RaviSharma1: Janet Yes I was only thinking of vision and radar sensors response to any obstruction, similarly other sensors to rain snow skidding
[13:01] Ram D. Sriram: @Ravi: I seem to have problems speaking into the system as I am in another meeting right now. My question is "Are these probability-based models scalable?"
[13:18] BobbinTeegarden: @john: holistic metaphor?
[13:25] RaviSharma1: Sargur many thanks for a wonderful talk on Probabilistic XAI.