15:04:54 <RRSAgent> RRSAgent has joined #webmachinelearning
15:04:54 <RRSAgent> logging to https://www.w3.org/2020/01/23-webmachinelearning-irc
15:04:59 <Zakim> Zakim has joined #webmachinelearning
15:05:04 <anssik> RRSAgent, make logs public
15:05:07 <Nikhil> Nikhil has joined #webmachinelearning
15:05:09 <anssik> Meeting: WebML CG Teleconference – 23 January 2020
15:05:14 <anssik> Chair: Anssi
15:05:19 <anssik> Agenda: https://github.com/webmachinelearning/meetings/blob/master/telcons/2020-01-23-agenda.md
15:05:40 <anssik> Scribe: Anssi
15:05:44 <anssik> scribeNick: anssik
15:05:46 <Chai> Chai has joined #webmachinelearning
15:05:54 <anssik> Present+ Anssi_Kostiainen
15:05:58 <anssik> Present+ Rafael_Cintron
15:06:03 <anssik> Present+ Ningxin_Hu
15:06:07 <anssik> Present+ Chai_Chaoweeraprasit
15:06:14 <anssik> Present- Ningxin_Hu
15:06:27 <anssik> Present+ Jonathan_Bingham
15:06:44 <anssik> RRSAgent, draft minutes v2
15:06:44 <RRSAgent> I have made the request to generate https://www.w3.org/2020/01/23-webmachinelearning-minutes.html anssik
15:06:49 <anssik> TOPIC: Announcements
15:07:00 <anssik> anssik: welcome our new WebNN API co-editor, Chai from Microsoft!
15:07:11 <anssik> -> https://github.com/webmachinelearning/webnn/pull/38 Add a co-editor
15:07:16 <jdarpinian> jdarpinian has joined #webmachinelearning
15:07:34 <anssik> anssik: Chai will work with Ningxin, help address issues, turn them into spec text, review PRs, triage issues
15:08:17 <anssik> TOPIC: Buffer sharing between GPU and ML accelerator
15:08:38 <anssik> anssik: let's sync on the status on VPU and DirectML prototyping
15:09:19 <anssik> https://github.com/webmachinelearning/webnn/issues/33
15:09:55 <anssik> chai: I think Ningxin now has enough data to move forward with POC, I provided him required details on ONNXRuntime
15:11:02 <anssik> ... issues with performance and VPU, think Ningxin should work together with VPU engineering folks, VPU is a bit tricky to get good performance
15:11:37 <anssik> ... also wanted to revisit discussion on using TF Lite as a platform for investigation
15:11:56 <anssik> https://www.w3.org/2020/01/09-webmachinelearning-minutes.html#x01
15:13:58 <Jonathan> Jonathan has joined #webmachinelearning
15:14:18 <anssik> nikhil: TF Lite makes sense, it's an abstraction layer that makes sense for this investigation as well
15:14:45 <anssik> Present+ Jonathan_Bingham
15:15:12 <anssik> TOPIC: conv2d native API mapping table - DirectML and ONNX
15:15:27 <anssik> anssik: thanks Chai for contributing the data for DML and ONNX
15:15:34 <anssik> -> https://github.com/webmachinelearning/webnn/blob/master/op_compatibility/conv2d.md conv2d op compatibility
15:16:07 <anssik> anssik: Chai please feel free to summarize your key findings, for details please see the table, what are the key implications?
15:16:17 <anssik> https://github.com/webmachinelearning/webnn/issues/28
15:17:06 <anssik> Chai: I was curious what's the thinking around grouped convolution? are they in scope?
15:17:44 <anssik> nikhil: we can make an argument to keep that as a separate op like TensorFlow does
15:18:57 <anssik> Chai: can define a new op or fold into existing op
15:19:14 <anssik> ... issue with combining is you'll have a big op hard to optimize
15:19:41 <anssik> ... if focus on compatibility, better to look at an area of overlap and see what works with all existing APIs, a more compelling argument
15:19:59 <anssik> ... also goes for conv3d
15:20:23 <anssik> anssik: hearing we want to keep these separate
15:21:10 <anssik> https://github.com/webmachinelearning/webnn/issues/28#issuecomment-555193911
15:23:02 <anssik> TOPIC: Float16 type support and handling unsupported OperandType
15:23:07 <anssik> anssik: resolution from our previous call was:
15:23:11 <anssik> ... Add "float16" to OperandType enum in WebNN API and define a way for the API to respond when float16 is not natively supported
15:23:27 <anssik> ... ningxin submitted a PR to update IDL
15:23:37 <anssik> -> https://github.com/webmachinelearning/webnn/pull/35 Add float16 and tensor-float16 into OperandType #35
15:23:47 <anssik> ... this PR only includes IDL changes, proposal to define handling of unsupported OperandType in a separate issue #36
15:23:55 <anssik> -> https://github.com/webmachinelearning/webnn/issues/36 Handling unsupported OperandType #36
15:24:08 <anssik> anssik: can we agree to merge PR #35 and address issue #36 separately, any concerns?
15:24:50 <anssik> anssik: [ hearning none ]
15:25:06 <anssik> anssik: Chai feel free to merge PR #35 after the call
15:25:15 <anssik> TOPIC: W3C Workshop on Web & Machine Learning
15:25:22 <anssik> anssik: please register by 21 Feb 2020
15:25:29 <anssik> -> https://www.w3.org/2020/01/machine-learning-workshop/ Workshop Call for participation
15:26:27 <anssik> anssik: committee expected to grow with 1-2 more members to increase diversity, good representation across industries already
15:27:54 <anssik> anssik: I started an agenda Google sheet and I'll share with the program committee soon
15:28:54 <anssik> anssik: sponsorship opportunities open
15:29:25 <anssik> -> https://www.w3.org/2020/01/machine-learning-workshop/sponsorship.html Sponsoring the W3C Workshop on Web & Machine Learning
15:29:50 <anssik> anssik: any questions? I encourage everyone to attend. Berlin is nice.
15:31:50 <anssik> TOPIC: Adjourn
15:32:05 <anssik> RRSAgent, draft minutes v2
15:32:05 <RRSAgent> I have made the request to generate https://www.w3.org/2020/01/23-webmachinelearning-minutes.html anssik
17:31:38 <Zakim> Zakim has left #webmachinelearning
18:30:57 <zkis> zkis has joined #webmachinelearning