Lalit, Per our call during our GSOC mentor check-in earlier this week, thanks for taking the lead in these standardization efforts and the formation of the working group.
James, thanks for your valuable feedback. For those interest in joining the working group around these efforts, we'll start out with a specific slack channel and can create a group on Discourse as well. To join the slack channel, visit https://join.slack.com/share/zt-fonegfmt-sPCtUok4vAXBYqzGdBWq9A As Lalit noted, we have several ongoing community efforts on the AI/ML front related to chatbots, credit analysis, computer vision for PPI, etc. as well as many more advanced partner-led innovations on this front hat we'd like to share the requirements and use cases for into the community. We welcome anybody that would like to advance these efforts to join the working group. Ed On Wed, Jul 8, 2020 at 2:13 PM James Dailey <jamespdai...@gmail.com> wrote: > Lalit +1 on this effort. > > Here is what I would add: > In looking at this area a couple of years ago, my company also came up > with some basic use case descriptions and some ideas. > > 1. Around credit scoring, it would be useful to be able to relate data > collected in the field (maybe held in a separate client intake database) > with larger datasets about demographics, economic conditions, other market > survey data, and historical call data records. > 2. For chatbots it would be useful to have a framework for issuing > queries to the account in a secure but consistent manner once the auth is > established > 3. Around additional use cases such as energy access programs or > agricultural insurance, it is useful to have historical spend data and > balances on the mifos/fineract accounts, but also historical data on > weather, energy grid infrastructure, ag commodity prices, and so on... > 4. The need for Federated learning across a network of nodes (where > organizations may not be able to share their data but could share their > insights) > > > So, I think that this effort should establish a set of Principles and then > a framework. > > - No Impact on Core Ops: Data Calls should not impede or create > additional calls on core Fineract/Mifos processing. > - Security and Privacy awareness/ compliance: Ensure that security > model and privacy concerns are addressed from the beginning > - APIs: There should be some standard APIs and ways to extend them; > likely there will be denormalized data structures (key pairs) ? > - Federated learning model and framework > - what else? > > Deployment matters - make it deployable and maintainable on some standard > toolsets ( Jupyter !) > Test data - for training models, some anonymized but realistic data would > be very useful. > > I'd also suggest that a way to run this as a separate microservice - > either against Fineract1.x or fineractCN would be a useful -future proof- > approach. Start perhaps with the Fineract1.x but follow the pattern found > on the Payment Gateway where the changes to fineract1.x were minimal and > the microservice could be built for fineractCN. I hope the repos on Mifos > around that should be clear. > > Thanks > > On Wed, Jul 8, 2020 at 11:13 AM Lalit Mohan S <slal...@gmail.com> wrote: > >> There are 3-4 projects leveraging AI/ML for use cases on credit scoring, >> chatbot, CV, etc. Each of these projects are using different ML packages, >> libraries, API calls. The idea is to standardize to 2-3 on-prem/on-cloud >> options so that deployment and management become easy. >> >> I hope it clarifies.. If not, we could have a call or elaborate your >> concern >> >> Regards >> Lalit >> >> On Wed, Jul 8, 2020 at 11:09 PM Saransh Sharma <sara...@muellners.com> >> wrote: >> >>> The idea of AI and ML platformification is too vague as of now. What >>> does it mean ? Could we get some kind of use case to start with ? >>> >>> This can be really tricky to build something like a standard layer where >>> data ingestion , and then pre-processing as well model training , I >>> don't think we should make fineract 1.x loaded with such heavy >>> intensive resource demanding processes. >>> >>> Fineract CN could be a nice approach towards building blocks since >>> services run in isolation. >>> >>> Focus should be on making the system real time. >>> >>> >>> On Wed, Jul 8, 2020 at 10:20 PM Abhijit Ramesh < >>> abhijitrames...@gmail.com> wrote: >>> >>>> Hi Lalit, >>>> >>>> As discussed in the meeting, I have been working with Deep Learning for >>>> around an year right now and am very interested in applying the knowledge I >>>> have to real world projects. >>>> I did go through the website that you have shared and have done some >>>> basic research on applying deep learning to Fintech especially for fraud >>>> detection seems to be really interesting. >>>> I am looking forward to being a part of the working group. >>>> >>>> >>>> Regards, >>>> Abhijit Ramesh >>>> <https://amfoss.in/> >>>> GitHub <https://github.com/abhijitramesh> | GitLab >>>> <https://gitlab.com/abhijitramesh> | Linkedin >>>> <https://www.linkedin.com/in/abhijit-ramesh-b583ab185/> >>>> >>>> >>>> >>>> >>>> On Wed, Jul 8, 2020 at 7:59 AM Lalit Mohan S <slal...@gmail.com> wrote: >>>> >>>>> Dear All, >>>>> >>>>> In GSOC Mentor call, we had a brief discussion on the need for >>>>> standardization of AI/ML options and packages considering the increasing >>>>> usage of AI/ML in Mifos/Fineract. >>>>> >>>>> There was a thought to start a working group to have a focused >>>>> discussion on the AI/ML standardization. >>>>> >>>>> Please share your inputs and interest to be part of the working group. >>>>> >>>>> Regards >>>>> Lalit >>>>> >>>>> -- >>>>> To unsubscribe from this group and stop receiving emails from it, send >>>>> an email to gsoc-mentors+unsubscr...@mifos.org. >>>>> >>>> -- >>>> To unsubscribe from this group and stop receiving emails from it, send >>>> an email to gsoc-mentors+unsubscr...@mifos.org. >>>> >>> >>> >>> -- >>> >>> Saransh Sharma >>> *Research Partner* >>> *Muellner Internet Pvt Ltd * >>> >>> >>> ---------------------------------------------------------------------------------------------- >>> >>> The idea of separation of me and you is amazing. >>> >>> ---------------------------------------------------------------------------------------------- >>> *Company Website <https://www.muellners.com/> **Company Linkedin >>> <https://www.linkedin.com/company/muellners> * >>> Company Facebook <https://www.facebook.com/muellners> >>> >>> This mail is governed by Muellner® Internet Private Limited's IT >>> policy. >>> The information contained in this e-mail and any accompanying documents >>> may contain information that is confidential or otherwise protected from >>> disclosure. If you are not the intended recipient of this message, or if >>> this message has been addressed to you in error, please immediately alert >>> the sender by reply e-mail and then delete this message, including any >>> attachments. Any dissemination, distribution or other use of the contents >>> of this message by anyone other than the intended recipient is strictly >>> prohibited. All messages sent to and from this e-mail address may be >>> monitored as permitted by applicable law and regulations to ensure >>> compliance with our internal policies and to protect our business. E-mails >>> are not secure and cannot be guaranteed to be error free as they can be >>> intercepted, amended, lost or destroyed, or contain viruses. You are deemed >>> to have accepted these risks if you communicate with us by e-mail. >>> >> Mifos-developer mailing list >> mifos-develo...@lists.sourceforge.net >> Unsubscribe or change settings at: >> https://lists.sourceforge.net/lists/listinfo/mifos-developer > > Mifos-developer mailing list > mifos-develo...@lists.sourceforge.net > Unsubscribe or change settings at: > https://lists.sourceforge.net/lists/listinfo/mifos-developer -- *Ed Cable* President/CEO, Mifos Initiative edca...@mifos.org | Skype: edcable | Mobile: +1.484.477.8649 *Collectively Creating a World of 3 Billion Maries | *http://mifos.org <http://facebook.com/mifos> <http://www.twitter.com/mifos>
_______________________________________________ Mifos-users mailing list Mifos-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/mifos-users
