Soft start; one of the discussants frames why the paper was chosen and what question we are testing it against.
Steel-man the paper. Walk the room through the contribution as the authors meant it to be read.
Critique. What does the paper not earn? What hidden assumption is doing the work? What would replicate, what would not?
The room argues. Newcomers and senior researchers in the same conversation; the chair keeps the thread tight.
What the room agreed on, what stayed contested, what we want to read next. Notes go on the mailing list within 48 hours.
Bi-weekly, Thursdays, 16:00 WAT. The specific Thursday and the paper go out on the mailing list one week ahead.
Hybrid. In-person at the AIRINA Labs office in Cotonou, Bénin. Remote attendees join by Zoom — link sent with the mailing-list announcement.
Bilingual EN / FR. Discussion switches between the two as the room requires; the chair will translate on request.
Read it if you can. Come anyway if you cannot — the discussants frame the paper at the top of the session, and the discussion is more useful for newcomers when they listen first.
Short discussion notes are sent on the mailing list within 48 hours of each session. The archive of past sessions and notes is shared with subscribers.
Free. The Paper Club is an open institutional activity; no registration fee in person or online.
The canonical introduction to TDA. Every subsequent paper assumes you have read this one — we open with it for exactly that reason.
The founding algorithm of persistent homology. Where the field was named and proven.
The most beautiful applied-TDA paper: a Klein bottle hiding in the patches of every photograph.
Mapper, the second pillar of TDA. Required before anything Mapper-shaped shows up in our own work.
The bridge from persistence diagrams to standard ML pipelines. Required before any downstream classifier reads a TDA feature.
The literature directly under our Phase 1 products. We agree, we disagree, we read it carefully.
The stability theorem. Why persistent homology is a real signal and not noise.
The other major vectorization of persistence diagrams. Bubenik vs. Adams, head to head.
The case for interpretable models, not post-hoc interpretations of black boxes. Why AIRINA products must be the former.
The high-stakes-medicine TDA paper. Precedent for our Phase 3 health work.
Get the next session's paper, date, and Zoom link a week ahead. Plus the discussion notes within 48 hours of each session. The list is low-volume: typically one announcement and one notes email per session.
Subscribe →Suggest a paper that fits the institute's lines: TDA, interpretable ML, applied mathematics for African systems. One paragraph on why it matters and what you would want the room to argue about. Cycle 3 is open for proposals.
Propose a paper →The Paper Club is in its first volume. Once Cycle 1 closes, the discussion notes, contested points, and follow-up references for each past session will be linked here — freely readable. Subscribers receive the notes by mail within 48 hours of each session in the meantime.