The Self-Organizing Conference on Machine Learning


Lunch at SOCML 2016

See the SOCML 2020 schedule here

SOCML ("SOC" pronounced like in social) is an informal gathering of machine learning researchers and practitioners focused on social interaction between participants. There are no publication proceedings and no talks with a presenter and an audience, only group discussions. SOCML makes diversity and inclusion a high priority. SOCML has a very small size cap so that it is possible for most participants to meet most other participants and the event takes place at "human scale". It is meant to some extent to recreate the atmosphere of chatting in the lobby outside the NeurIPS deep learning workshop circa 2010. It is a great venue for getting the latest information (before it has had the time to pass through peer review) and to share ideas that are not quite baked enough to fit into the traditional academic mold.

SOCML has always been Ian Goodfellow's personal project, but carried out with support from many people over the years, including especially Kate Miltenberger, Negar Saei, and Rainier Aliment. SOCML first took place in 2016 at OpenAI's support in San Francisco, then took place with Google's support in Sunnyvale in 2017 and Toronto in 2018. There was no SOCML 2019 because Ian was too busy to run it after welcoming a new baby to the family. SOCML has always been free to all participants. In 2016-2018, SOCML was an in-person event, complete with a venue, travel grants, food, after parties and so on. In 2020, due to the ongoing COVID pandemic, SOCML will be 100% online.