General Assembly 2023 - Roadmap 2024

[thibaultcordier]

Join us for our upcoming General Assembly on Friday, November 24th, 2023, 17pm (UTC+1), which will take place at Quantmetry's offices, 52 rue d'Anjou - 75008 PARIS (limited number of places on site) with a videoconference. We have some exciting developments to share!

🚀 What's on the Agenda?

🔹 MAPIE's Priority Contributions for 2024: Join us as we unveil the roadmap for our next major release. We'll dive deep into our focus areas, with a special emphasis on Adaptive Uncertainty Quantification and Conformal Testing.

🗣️ Your Voice Matters: Your insights and opinions are invaluable to us. This meeting offers a unique opportunity to share your thoughts on our upcoming plans, as well as any other features and issues you'd like to address. We believe in collaborative growth, and your input shapes the future of MAPIE.

📝 Registration Details: Make sure you register for the event using our form to ensure you are kept up to date with the latest information (meeting connection link and venue, etc). Don't miss out on this chance to connect, engage, and have an impact on MAPIE (https://forms.office.com/e/xwrd5Q7UaT).

Stay tuned for more updates as we approach the date. See you at the General Assembly on November 24th! 🗓️

[thibaultcordier]

We wish you all the best for 2024! And to kick off the new year with some good news, here's our development and research avenues presented at the General Meeting we organised last month MAPIE_General_Meeting_2023.pdf.

Now, we're taking this opportunity to make official the 2024 roadmap that we intend to follow:

Main Focus - P1

• Mondrian Conformal Prediction: validity within categories thanks to Mondrian approach for both regression and classification
• Adaptive Conformal Prediction: approaches based on non-conformity scores distribution estimation
• Hypothesis Testing: toolbox of hypothesis tests for exchangeability and performance drift
• Risk control: selective prediction for regression and classification with Learn-Then-Test framework

Priorities - P2

• Interoperability: support natively pytorch, tensorflow, transformers, etc.
• Large Scale Deployment: accelerate code, be robust to memory problems...
• Use Cases: instance segmentation, object detection, generation (tabular data, text), etc.

Many of you have already contributed to MAPIE earlier this year, and we're delighted to be able to involve you in this project.

Thank you very much and see you soon!