First NeurIPS conference

Background

NeurIPS, the Conference on Neural Information Processing Systems, began in 1987 ^[1]. The meeting was first proposed a year earlier, in 1986, at the annual invitation-only Snowbird Meeting on Neural Networks for Computing, which was organised by Caltech and Bell Laboratories ^[1]. That earlier gathering was held by invitation; the proposal that came out of it was for a conference with its own published record of accepted work.

The first meeting

The first NeurIPS conference was held in 1987, with the information theorist Ed Posner as conference president and the learning theorist Yaser Abu-Mostafa as program chairman ^[1]. The conference was sponsored by the IEEE ^[1]. It took place in Denver, Colorado, and Denver remained its location from 1987 until 2000 ^[1].

What the programme looked like

The conference was built around a programme of three days of invited talks, together with oral and poster presentations of refereed papers — that is, papers that had been read and assessed by other researchers before being accepted ^[1]. Those three days were followed by two further days given over to workshops and competitions ^[1].

The proceedings

The first proceedings was published in book form in 1987 by the American Institute of Physics, under the title *Neural Information Processing Systems* ^[1]. The official NeurIPS proceedings archive lists the same volume as *Neural Information Processing Systems 0, NIPS 1987*, edited by D. Anderson ^[2]. Among the papers collected in it were *Synchronization in Neural Nets*, *How Neural Nets Work* and *Strategies for Teaching Layered Networks Classification Tasks* ^[2].

The volume numbering that started with that book has been kept ever since. The NeurIPS proceedings index lists volume 0 as NIPS 1987 and volume 1 as NIPS 1988 ^[3], and the same index lists *Advances in Neural Information Processing Systems 38* as NeurIPS 2025 ^[3]. The book published after the first meeting is therefore the beginning of a numbered series that still runs today.

Why it mattered

Wikipedia identifies Ed Posner as the founder of NeurIPS, and says that he established the NeurIPS Foundation, the body that organised later conferences ^[1]. The first meeting thus set up more than a one-off event: it produced a continuing series of proceedings volumes ^[3] and an organisational structure that carried the conference forward ^[1]. The naming in the archive also preserves the change of label — the earliest volumes are recorded as NIPS 1987 and NIPS 1988 ^[3], while the conference and its series are now known under the NeurIPS name ^[1].