sg14 at CppCon is an interesting idea. If that's the plan, please tell us in advance such that I ask Jon not to schedule my talk(s) the same day.


Le ven. 10 juin 2022 à 11:44, Michael Wong via SG19 <> a écrit :

On Wed, Jun 8, 2022 at 3:15 PM Michael Wong <> wrote:
Hi all, SG19 Machine Learning meeting will focus on graphs. Michael Wong is inviting you to a scheduled Zoom meeting.

Topic: SG19 monthly
Time: 02:00 PM Eastern Time (US and Canada)
    Every month on the Second Thu,

Join from PC, Mac, Linux, iOS or Android:
    Password: 035530

Or iPhone one-tap :
    US: +13017158592,,93084591725# or +13126266799,,93084591725#
Or Telephone:
    Dial(for higher quality, dial a number based on your current location):
        US: +1 301 715 8592 or +1 312 626 6799 or +1 346 248 7799 or +1
408 638 0968 or +1 646 876 9923 or +1 669 900 6833 or +1 253 215 8782
 or 877 853 5247 (Toll Free)
    Meeting ID: 930 8459 1725
    Password: 035530
    International numbers available:

Or Skype for Business (Lync):


1. Opening and introductions

The ISO Code of conduct:

IEC Code of Conduct:

ISO patent policy.

The WG21 Practices and Procedures and Code of Conduct:

1.1 Roll call of participants
Chris Ryan, Jens Maurer, Kevin Dewesee, Ozan Irsoy, Phil Ratzloff, Rene Rivera, Richard Dosselment, Scott McMillan, MIchael Wong, ka ming chan, Will wray

1.2 Adopt agenda

1.3 Approve minutes from previous meeting, and approve publishing
 previously approved minutes to

1.4 Action items from previous meetings

2. Main issues (125 min)

2.1 General logistics

Meeting plan, focus on one paper per meeting but does not preclude other
paper updates:

    June 9, 2022 02:00 PM ET: Graph
    Jul 14, 2022 02:00 PM ET: Matrix, RL and DC
    Aug 11, 2022 02:00 PM ET: Stats
    Sep 13, 2022 02:00 PM ET: Graph
    Oct 12, 2022 02:00 PM ET: Matrix RL/DC

ISO meeting status

future C++ Std meetings     The submissions deadline is June 24th, with decisions sent by July 31st.

2.2 Paper reviews

2.2.1: ML topics Graph Proposal Phil Ratsloff et al

Latest paper:

Here’s a link to the paper (different than the previous paper reviewed).
There are some additional updates I’m planning on making before the meeting.

edgeless view
2nd set of functions are for edges on a specific vertex
not supporting due to pairing down so it is returning incoming edges as well
neighbors now instead of adjacency and get source_id and target_id
undirected incidence graph
concept bounds polymorphism on an algo, or like an abstract base class: the undirected incidence and directed incidence graph, the algo used should be well defined - when algo gets target_id, in an undirected graph, it goes to a specific value which is well-defined, but the physical storage can be element 0 source, and 1 is the target for directed graph, but undirected element 0 or 1 could be source or target

undirected graph needs a source-incidence graph (both a source and target id on an edge), directed vs undirected through emulation?
do we need to store source id? 
undirected edge value 
undirected graph needs a source and target since we need to know where we came from

different names for directed vs undirected or should be unordered on the edge
terminology should reflect the math
better term? unordered edge/ vertices, as long as uid and vid comes out in the result

what's returned by edge list? edgeless G, E
can now port NW graph algorithms as interface is very similar

SG14 at cppcon?

More simple stats:  added missing stats: median and mode because they are more involved
weighted percentile, may not have weighted variant of this
looked at python which has bivariate values
might try to extend what we have to support these multivariate stats to be on par with python
prediction needs regression
return a start and end iterator
JM: template argument of subrange can be omitted as it is defaulted 
what if looking for n modes in n range: allow user to specify how many modes
JM: why do those 2 ranges need to be same type: value type needs to be friends; yes will make separate
might also need constraints on relationship of value type and allow arithmetic
dof is passed 1 and it will subtract just like python

look at newer ranges algo do

accumulate might need a zipview if precondition has the same length

instead of univariate, bivariate we can condense that down to one

range end iterator is always exclusive : it points to one past the last thing and not second element
mode is sorted Reinforcement Learning Larry Lewis Jorge Silva

Reinforcement Learning proposal: Differential Calculus: Stats paper

Current github

Stats review Richard Dosselman et al

Feedback from Johan Lundberg and Oleksandr Korval

P1708R3: Math proposal for Machine Learning: 3rd review

PXXXX: combinatorics: 1st Review

*> above is the stats paper that was reviewed in Prague*
*> Review Jolanta Polish feedback.*
<>* Matrix paper

2.2.3 any other proposal for reviews?

2.3 Other Papers and proposals

P1416R1: SG19 - Linear Algebra for Data Science and Machine Learning

P1415: Machine Learning Layered list

2.2.2 SG14 Linear Algebra progress:
Different layers of proposal

2.5 Future F2F meetings:

2.6 future C++ Standard meetings:


3. Any other business

New reflector

Old Reflector!newtopic/sg19

Code and proposal Staging area

4. Review

4.1 Review and approve resolutions and issues [e.g., changes to SG's
working draft]

4.2 Review action items (5 min)

5. Closing process

5.1 Establish next agenda

5.2 Future meeting

   May 12, 2022 02:00 PM ET: Stats
    June 9, 2022 02:00 PM ET: Graph
    Jul 14, 2022 02:00 PM ET: Matrix, RL and DC
    Aug 11, 2022 02:00 PM ET: Stats
    Sep 13, 2022 02:00 PM ET: Graph
    Oct 12, 2022 02:00 PM ET: Matrix RL/DC
SG19 mailing list