15:01:03 RRSAgent has joined #webmachinelearning 15:01:03 logging to https://www.w3.org/2021/02/18-webmachinelearning-irc 15:01:06 RRSAgent, make logs Public 15:01:08 please title this meeting ("meeting: ..."), anssik 15:01:09 Meeting: WebML CG Teleconference – 18 February 2021 15:01:14 Chair: Anssi 15:01:20 Agenda: https://github.com/webmachinelearning/meetings/blob/master/telcons/2021-02-18-agenda.md 15:01:24 Present+ Ningxin_Hu 15:01:24 Scribe: Anssi 15:01:28 scribeNick: anssik 15:01:35 Present+ Anssi_Kostiainen 15:01:39 dom has joined #webmachinelearning 15:01:39 Present+ Dominique_Hazael-Massieux 15:01:55 Present+ Rafael_Cintron 15:02:09 Present+ Chai_Chaoweeraprasit 15:02:52 RRSAgent, draft minutes v2 15:02:52 I have made the request to generate https://www.w3.org/2021/02/18-webmachinelearning-minutes.html anssik 15:03:02 Taimur has joined #webmachinelearning 15:03:14 Topic: WG Charter update 15:03:37 Present+ Geun_Hyung_Kim 15:03:55 GeunHyung_Kim has joined #webmachinelearning 15:04:04 Present+ 15:04:05 anssik: WG Charter update and next steps from W3C Staff Dom 15:04:29 Dom: first, sorry this has taken longer than I had initially hoped 15:04:45 RafaelCintron has joined #webmachinelearning 15:04:51 ... by next week Wed-Thu the draft Charter for the WG which has been tweaked should go by the W3C AC review 15:05:00 ... reps from all W3C Members 15:05:12 ... they have provide feedback and vote 15:05:43 ... most importantly that is an opportunity for the reps to comment if they have change proposals, in most cases the changes are around the scope of the charter 15:05:54 ... charter delineates IPR commitments members do while joining the group 15:06:01 -> https://w3c.github.io/machine-learning-charter/charter.html Web Machine Learning Working Group Charter 15:06:15 Dom: also privacy and accessibility perspective reviewed 15:06:26 ... 4 weeks review time for all W3C Members 15:06:45 ... also part of this process, demonstrating there's broad support for starting standardization 15:07:07 ... we expect 5% or more of W3C Membership to express support, that is 20-22 distinct W3C Members 15:07:26 ... it is really important you get your W3C Advisory Committee reps sent a supportive review of the charter 15:07:48 ... and encourage your colleagues in others organizations to also get their AC reps pay attention to the review and support the work 15:08:26 ... I will be also working contacting various companies who have expressed interest in this space, ML is pretty exciting topic so many people should be paying attention 15:09:25 Dom: after 4 weeks AC review, if there has been no so-called Format Objections, and we get clear signals of support the charter gets approved in a few days and the WG gets started end of March to early April 15:09:47 ... if Formal Objections are filed, then resolving them is an involved process 15:10:07 ... to sum this up, if things go smoothly we should be able launch end of March early April 15:10:22 ... when we're a WG, you are asked to join the WG with your W3C Member affiliation 15:10:49 ... there's a path to become an Invited Experts for W3C non-Members 15:11:34 ... difference CG vs. WG, you'll get W3C Staff contact for the WG to help with W3C Process, best practices from W3C groups, consistent with other W3C efforts 15:12:18 ... all the time, once the group has gathered enough members, we start more formal steps, such as First Public Working Draft publication with Royalty-Free licensing commitments 15:15:20 Present+ Sandeep_Gupta 15:15:51 Topic: TAG review feedback 15:16:06 anssik: Discuss key TAG review issues per GH feedback and activity 15:16:14 ... note GH label change from "tag" to "tag-tracker" 15:16:21 -> https://github.com/webmachinelearning/webnn/issues?q=is%3Aissue+is%3Aopen+label%3Atag-tracker All GH issues with [tag-tracker] label 15:16:34 anssik: Let's discuss those first that have received comments. 15:16:48 Subtopic: [tag-tracker] NamedOutput mechanism clarification 15:16:54 -> https://github.com/webmachinelearning/webnn/issues/140 NamedOutput mechanism clarification #140 15:17:21 anssik: Ningxin says, basically, the NamedOutputs is just used to index an Output by string-based name. 15:17:31 anssik: specific text to add to the spec to clarify? 15:19:57 Dom: horizontal review is increasingly more important when advancing to WG, one of those groups is W3C Technical Arch Group that reviews APIs for architectural aspects 15:21:22 ... Privacy Interest Group (PING) is another such horizontal review group, others are Accessibility review, Security review, and Internationalization review, not expecting much I18N or A11Y review 15:21:35 ... Privacy is likely more interested in this work and has substantial feedback 15:22:12 ... "w3cbot" is a helper that keeps track of this horizontal review and makes sure the feedback is appropriately addressed 15:22:47 ... "w3cbot" will help demonstrate wide review feedback has acted upon in a proper manner 15:23:48 Subtopic: [tag-tracker] String enum for activations 15:23:55 -> https://github.com/webmachinelearning/webnn/issues/138 String enum for activations #138 15:25:19 Chai: AFAIK, the feedback is to create a separate enum for activations used elsewhere in the API, the issue in this is that in recurrent networks e.g. GRU different activations may be used 15:25:46 ... not all can be used in RNNs, so more appropriate to have a blanket enum for all activations, because alternative solution needs a runtime check 15:26:16 ... then the API would need to reject that, so preference would be to define an API in such a manner it does not fail easily in common cases 15:26:44 ... in this sense, having a specific enum for activations prevents that, i.e. if a new version adds new activations we can just add an additional value to the enum 15:28:01 Chai: we could have a QA section for this spec and park these issues these? 15:28:53 https://tabatkins.github.io/bikeshed/ 15:28:58 https://tabatkins.github.io/bikeshed/#metadata-inline-github-issues 15:29:11 https://tabatkins.github.io/bikeshed/#remote-issues 15:29:46 Subtopic: [tag-tracker] Constructor or builder pattern for model building? 15:29:54 -> https://github.com/webmachinelearning/webnn/issues/136 Constructor or builder pattern for model building? #136 15:31:26 Chai: factory is better if you'd have multiple model builders in the future 15:32:13 ... you don't want to expose a constructible type, but have a factory method, .createXXX rather than new FoobarXXX() 15:32:23 ... eager mode would require different builders 15:32:37 ... so the API is probably better stick with the factory method 15:33:57 Dom: I don't know of many Web APIs that use factory for object construction, re future-proofing the API, I'm wondering how this envisioned on the web where we don't deprecate APIs and APIs can evolve over time, overloading constructors, hiding. 15:34:14 ... The reason for TAG response is that factory is not a common pattern in Web APIs 15:34:54 Chai: that is interesting, I'm coming from OS API versioning point of view and design patterns in that context 15:35:27 ... in OS platform sense, unless we're really certain there's no other way to construct an object, we'd prefer to have an abstract way to create an object 15:36:28 Dom: this is usual design pattern in other context, the message the TAG sends here is consistency 15:37:29 https://w3ctag.github.io/design-principles/ 15:37:43 https://w3ctag.github.io/design-principles/#constructors 15:41:25 Subtopic: [tag-tracker] getNeuralNetworkContext() and createModelBuilder() params 15:41:34 -> https://github.com/webmachinelearning/webnn/issues/135 getNeuralNetworkContext() and createModelBuilder() params #135 15:41:54 anssik: Chai makes a point about the context vs. the model builder 15:42:33 Chai: this seems to be a question from TAG, asking whether these two methods have the right parametrization 15:42:53 ... from the conceptual point of view, we want to have two objects constructed by the implementer of this interface 15:43:21 ... one is the context object that holds the global state, e.g. NPU or GPU device 15:43:37 ... another object is the model builder that holds transient state specific to the topology constructed 15:43:45 ... that's the difference between these two things 15:44:08 ... the response explains why we need both 15:44:53 Chai: the question is perhaps not specific enough? 15:45:19 Dom: maybe have a call with Sangwhan so this can be synchronously discussed (timezone issue) 15:47:49 Subtopic: [tag-tracker] Training in the batch normalization section 15:47:57 -> https://github.com/webmachinelearning/webnn/issues/134 Training in the batch normalization section #134 15:48:04 Discussed in the catch-all TAG review issue: https://github.com/w3ctag/design-reviews/issues/570#issuecomment-774559266 15:49:05 https://github.com/webmachinelearning/webnn/pull/144 15:49:41 Subtopic: [tag-tracker] issues without comments 15:49:46 -> https://github.com/webmachinelearning/webnn/issues?q=is%3Aissue+is%3Aopen+label%3Atag-tracker+comments%3A%3C1 All GH issues with [tag-tracker] label and no comments yet 15:50:45 anssik: any questions re TAG feedback? 15:51:00 Subtopic: Explainer feedback 15:51:04 -> https://github.com/w3ctag/design-reviews/issues/570#issuecomment-768151707 Explainer feedback part 1 15:51:09 -> https://github.com/w3ctag/design-reviews/issues/570#issuecomment-768861645 Explainer feedback part 2 15:51:33 anssik: I can take a look, other volunteers welcome. 15:52:06 Dom: +1 to make this use case driven 15:52:17 ... I'll be off next week, otherwise happy to help 15:53:57 Topic: Privacy IG feedback and the Security and Privacy self-review 15:54:03 anssik: Wanted to discuss the early W3C Privacy Interest Group (PING) feedback and update PR #132 accordingly 15:54:25 -> https://github.com/webmachinelearning/webnn/issues/119#issuecomment-772565085 PING early privacy review 15:54:31 anssik: I'd like to update PR #132 with changes from PING, if any 15:54:39 anssik: PING feedback 1: "What do you think would be appropriate to prevent, deter or minimise sites from misusing or abusing the capabilities of this API? (Note: This is not a problem unique to this API, and perhaps solutions discovered here could help fix problems in JavaScript, WebGL, etc.)" 15:55:43 q+ 15:55:48 ack RafaelCintron 15:56:14 Sandeep has joined #webmachinelearning 15:56:29 anssik: PING feedback 3: "Is the API restricted to first-party contexts? Or do third-party frames have access? (The answer to 2.13 of the Self-Review: Security and Privacy Questionnaire (above) suggests they do, and that you are exploring the potential of a policy-controlled feature approach.) Is there any reason not to simply restrict to first party context? (i.e. what are the likely use cases you envision that would require 15:56:29 third-party frames to have access to the API?)" 15:57:02 RafaelCintron: I'm supportive of top-level doc needing to grant access to iframes 15:57:35 ... PING feedback 1 hard to answer, because many of these powerful APIs on the platform already let do the same things 15:59:52 anssik: PING feedback 2: "A related question - suppose the API was enabling machine learning use cases involving mouse movements (an example of behavioural biometrics), what are your thoughts about user awareness/consent and mitigations for abuse?" 16:02:35 anssik: ran out of time, PING feedback 4-7 deferred to future call 16:02:41 Topic: Adjourn 16:02:49 RRSAgent, draft minutes v2 16:02:49 I have made the request to generate https://www.w3.org/2021/02/18-webmachinelearning-minutes.html anssik 18:00:21 Zakim has left #webmachinelearning