April 26–30, 2027 Singapore, SGSuntec Singapore Convention Centre
annual
ICLR is one of the premier venues for deep learning and representation learning research, run as a double-blind peer-reviewed conference with a notably competitive acceptance process. The 2026 edition in Rio de Janeiro saw submissions surge 68 percent year-over-year, while in-person attendance tightened around a more selective accepted-paper crowd.
Who goesML researchers, PhD students, AI lab scientists and recruiters, and academic faculty
ICLR consistently delivers high-signal research content across representation learning, reasoning, and agents, and the 2026 edition in Rio was praised for the quality of researcher interactions despite a drop in overall in-person attendance. The shift in research focus from foundation model refinement toward reasoning and agentic behavior made the accepted-paper slate particularly valuable. The main friction points were venue logistics and a notable security breach in the review system that exposed author and reviewer identities for a large submission subset.
What works
+Acceptance rate dropped to 27 percent in 2026 (from 32 percent in 2025), meaning the accepted paper track represents genuinely high-bar research
+Attendees consistently report it is easy to meet top researchers across subfields in an informal setting, with strong hallway-track density
+Workshops program is broad and deep, spanning emerging subfields before they reach the main track
What doesn't
−A 2026 security incident exposed author, reviewer, and area chair identities via the OpenReview API, raising ongoing concerns about review integrity
−In-person attendance fell sharply in 2026 (7,054 vs 10,435 in 2025 Singapore), limiting spontaneous networking volume relative to prior years
Attended ICLR? Rate it for 7 days of Pro Mini.
Honest verdict in 60 seconds. Trial activates the moment you submit.
Attendee reviews (0)
No attendee reviews yet. Be the first — write a quick verdict and we'll grant 7 days of Pro Mini.
