14:29:38 <RRSAgent> RRSAgent has joined #webmachinelearning
14:29:38 <RRSAgent> logging to https://www.w3.org/2019/01/10-webmachinelearning-irc
14:29:44 <Zakim> Zakim has joined #webmachinelearning
14:29:48 <anssik> RRSAgent, make logs public
14:29:51 <Ningxin_Hu> Ningxin_Hu has joined #webmachinelearning
14:29:52 <anssik> Meeting: WebML CG Teleconference – 10 January 2019
14:29:56 <anssik> Chair: Anssi
14:30:02 <anssik> Scribe: Anssi
14:30:08 <anssik> Agenda: https://github.com/webmachinelearning/meetings/blob/master/telcons/2019-01-10-agenda.md
14:30:12 <anssik> scribeNick: anssik
14:30:18 <kreeger> kreeger has joined #webmachinelearning
14:30:19 <anssik> Present+ Anssi_Kostiainen
14:30:23 <anssik> Present+ Ningxin_Hu
14:30:25 <nsthorat> nsthorat has joined #webmachinelearning
14:30:31 <anssik> Present+ Tomoyuki_Shimizu
14:30:33 <chrisheilmann> chrisheilmann has joined #webmachinelearning
14:30:45 <kreeger> + Nick Kreeger
14:30:51 <anssik> Present+ Nick_Kreeger
14:30:51 <RafaelCintron> Present+ RafaelCintron
14:30:56 <nsthorat> Present+ Nikhil Thorat
14:31:02 <anssik> Present+ Nikhil_Thorat
14:31:17 <chrisheilmann> Present + chrisheilmann
14:31:38 <anssik> Present+ Rafael_Cintron
14:31:40 <belem> belem has joined #webmachinelearning
14:31:56 <anssik> Present+ Daniel_Mazurkiewicz
14:32:27 <anssik> RRSAgent, draft minutes v2
14:32:27 <RRSAgent> I have made the request to generate https://www.w3.org/2019/01/10-webmachinelearning-minutes.html anssik
14:32:34 <anssik> TOPIC: Welcome
14:32:42 <anssik> anssik: Welcome back! We've kicked off 2019 with good discussion and new members e.g. from Google, Brain.js community.
14:33:09 <anssik> TOPIC: High level vs low level follow-up
14:33:19 <anssik> -> High level vs low level follow-up https://github.com/webmachinelearning/webnn/issues/3
14:33:28 <anssik> anssik: Thanks for active discussion and proposals in the high vs low level issue
14:33:42 <anssik> ... It is important we find alignment in the group around this topic before we advance to concrete API spec work
14:33:50 <anssik> anssik: Goal today to review & solicit feedback on the existing proposals, see if we can agree on the next steps
14:33:59 <anssik> ... Identified 3 proposals to start, open to more suggestions:
14:34:04 <anssik> ... a) Incubate both the operator and load model approaches in parallel
14:34:18 <anssik> ... b) Focus on the operator approach and develop the load model approach as a polyfill initially
14:34:35 <anssik> ... c) New HTML element similar to <video>
14:34:50 <anssik> ... Also please note recently contributed "WebAI" proposal by a Brain.js community member
14:35:08 <anssik> anssik: Option b) would not require changes to the charter, options a) and c) might require a charter amendment (we have a process for that)
14:35:37 <anssik> anssik: I'd like to open the floor for comments.
14:35:46 <anssik> q+ would like to say something
14:36:01 <gregwhitworth> present+
14:36:07 <gregwhitworth> q+
14:36:13 <anssik> ack gregwhitworth
14:36:24 <dsmilkov> dsmilkov has joined #webmachinelearning
14:38:15 <anssik> [gregwhitworth summarizing his proposal at https://github.com/webmachinelearning/webnn/issues/3#issuecomment-447080045]
14:39:40 <anssik> gregwhitworth: would prefer option a)
14:40:00 <anssik> chrisheilmann: would also support option a)
14:41:09 <anssik> danielm: How quickly we expect to develop the standard? Can we split the standard development in phases?
14:45:12 <anssik> kreeger: from Google side prefer operators approach, i.e. option b)
14:46:07 <nsthorat> WebAI proposal: https://github.com/DanielMazurkiewicz/WebAI
14:46:49 <anssik> [danielm introducing Brain.js]
14:47:08 <kreeger> Google opinion: Prefer ops in this proposal over model loading. Model architecture and format is very fluid and continues to evolve. We think it makes sense to wait until model formats and more flushed out before proposing a format.
14:47:31 <anssik> RRSAgent, draft minutes v2
14:47:31 <RRSAgent> I have made the request to generate https://www.w3.org/2019/01/10-webmachinelearning-minutes.html anssik
14:48:45 <anssik> anssik: Could you use the operator-based WebNN API as an alternative backend for Brain.js?
14:51:11 <anssik> danielm: yes, from technical perspective
14:52:44 <anssik> anssik: hearing we need to continue discussion in the issue toward concensus
14:52:44 <kreeger> I think more debate vs. a and b
14:52:59 <anssik> TOPIC: Web Neural Network API - requirements
14:53:05 <anssik> -> Web Neural Network API https://webmachinelearning.github.io/webnn/
14:53:12 <anssik> anssik: Use cases landed, thank you to all contributors!
14:53:19 <anssik> -> Use cases https://webmachinelearning.github.io/webnn/#usecases
14:53:32 <anssik> anssik: proposed next step to derive requirements to measure API against
14:53:38 <anssik> -> [Example] WebRTC requirements derived from use cases https://w3c.github.io/webrtc-nv-use-cases/
14:53:49 <anssik> anssik: Do people feel a list of concrete requirements would help us make more informed API design decisions?
14:54:29 <anssik> anssik: Support to derive requirements?
14:54:55 <anssik> chrisheilmann: I'd prefer to do that, also look into new use cases. Do more outreach.
14:55:33 <anssik> PROPOSED RESOLUTION: Add requirements derived from use cases to the spec
14:56:14 <daniel> https://github.com/BrainJS/brain.js
14:56:33 <anssik> RESOLUTION: Add requirements derived from use cases to the spec, and add more use cases
14:56:44 <anssik> TOPIC: Web Neural Network API - semantic segmentation investigation
14:56:54 <Ningxin_Hu> semantic segmentation use case: https://webmachinelearning.github.io/webnn/#usecase-segmentation
14:56:56 <anssik> anssik: Ningxin has an update on semantic segmentation use case investigation
14:57:02 <anssik> -> WebNN Use Case Investigation: Semantic Segmentation slides https://docs.google.com/presentation/d/1FYiirLCQgAZbGLxm9zHDuRSsSFZpvUmwUVYQqdOpQBM/
14:57:32 <anssik> [Ningxin discusses material on slides: use case, example, identified requirements of DeepLabV3+ MobileNet V2 model]
14:58:09 <gabe> gabe has joined #webmachinelearning
14:59:17 <anssik> RRSAgent, draft minutes v2
14:59:17 <RRSAgent> I have made the request to generate https://www.w3.org/2019/01/10-webmachinelearning-minutes.html anssik
15:01:45 <jawache> jawache has joined #webmachinelearning
15:06:56 <gregwhitworth> custom operator capability
15:08:59 <anssik> anssik: a good to have a new use case and requirement for custom operators
15:10:50 <anssik> tomoyuki: We should discuss custom operator proposal more, how to implement it. What is the coverage of operators?
15:11:15 <anssik> TOPIC: Web Neural Network API - call for editors
15:11:31 <anssik> anssik: Call for editors. I'd like to see if there are volunteers to help edit the spec(s) in this group.
15:11:44 <anssik> ... Simply put, the responsibility of an editor is to translate group's consensus position into spec text, resolve and triage issues
15:12:43 <anssik> TOPIC: Any Other Business
15:12:52 <anssik> anssik: Last month I shared with you a plan for a proposed W3C workshop.
15:13:10 <anssik> ... After discussion with key stakeholders, decided to postpone the workshop from June to October 2019.
15:13:31 <anssik> -> https://www.w3.org/2018/11/machine-learning-workshop/ [DRAFT] W3C Workshop on Web & Machine Learning
15:13:54 <anssik> ... The workshop draft is open to further feedback.
15:15:27 <anssik> danielm: If we'd like to include training in scope, how to do that?
15:16:42 <dsmilkov> Thanks everyone and thank you Anssi for leading this meeting!
15:16:46 <anssik> TOPIC: Adjourn
15:16:48 <belem> belem has left #webmachinelearning
15:16:51 <anssik> RRSAgent, draft minutes v2
15:16:51 <RRSAgent> I have made the request to generate https://www.w3.org/2019/01/10-webmachinelearning-minutes.html anssik
15:20:24 <anssik> [Re scope expansion, process for charter amendment https://webmachinelearning.github.io/charter/#charter-change asks for 2/3 votes to pass.]
15:20:26 <anssik> RRSAgent, draft minutes v2
15:20:26 <RRSAgent> I have made the request to generate https://www.w3.org/2019/01/10-webmachinelearning-minutes.html anssik
17:00:07 <Zakim> Zakim has left #webmachinelearning