SALT #3 - Theory-focused ML paper reading group

Tired of attending reading groups? You read papers, but ideas don't stick? You want to go deeper but procrastinate anyway? We're organizing SAFL Learning and Theory Club or SALT for short - a theory-heavy ML research paper reading session where we sit down and work through a proof. The goal is not to sprint through papers but to take a single theorem, lemma, or idea and try to understand it deeply. So, if pondering at the whiteboard is your kinda thing, then you'll feel right at home at our reading group.

What are we reading this week? Continue reading Bergstr ̈aßer et. al. "Transformers are inherently succinct" (ICLR 2026 Outstanding Paper) (Formal methods and TCS esp. complexity theory pre-reqs) Read Aaronson's essay on "Why Philosophers Should Care About Computational Complexity" (Basic TCS, Complexity theory pre-req) Read Conitzer et. al. "Fundamental Impossibility Results in AI Safety: Trading Off between Desiderata" (Game theory pre-req likely) We will take a group vote and read the most voted paper. What to expect? 10 mins of context-setting, 30 mins of derivations by a facilitator, followed by discussions and Q&A. Light refreshments included! Where? Secure AI Hub, Bengaluru When? Nov 14th, 4pm onwards

RSVP sooner; we only ask for enough information to curate a small, intentional group. Bring nothing but your curiosity. See you there! Hosted by Secure AI Futures Lab (SAFL)

---

Curated listing from Luma. Organized by Secure AI Futures Lab | SAFL. RSVP happens on the organizer's page until they claim it. Run this event? Claim it on Jamao to get free RSVPs, WhatsApp reminders and UPI payouts.