ICML 2020

ICML 2020
Name	ICML 2020
Genre	Academic conference
Status	Completed
Frequency	Annual
Location	Virtual (originally planned for Vienna)
Years active	1980–present
First	1980
Organizer	International Machine Learning Society

ICML 2020 was the 37th annual meeting of the International Conference on Machine Learning, held in a virtual format after plans for a physical venue in Vienna were altered. The conference brought together researchers from institutions such as Google Research, Microsoft Research, DeepMind, Facebook AI Research, and OpenAI, alongside universities like Stanford University, Massachusetts Institute of Technology, University of Toronto, Carnegie Mellon University, and University of Oxford. Major sponsors and stakeholders included NIPS Foundation-adjacent companies and associations such as the Association for Computing Machinery and contributors from industry labs like Amazon Web Services and NVIDIA.

Overview

ICML 2020 continued the lineage begun with the founding community connected to the Neural Information Processing Systems tradition and the International Machine Learning Society. The virtual meeting echoed developments seen at other major gatherings such as CVPR, ACL, ICLR, KDD, and SIGIR, emphasizing advances in deep learning, probabilistic modeling, reinforcement learning, and optimization. Participants included representatives from research groups led by scholars affiliated with Yann LeCun, Geoffrey Hinton, Yoshua Bengio, Andrew Ng, and institutions like Berkeley AI Research and Caltech. The program reflected cross-pollination with efforts from labs associated with IBM Research, Bell Labs, and Huawei Noah's Ark Lab.

Organization and Venue

Originally scheduled for Vienna, the organizing committee had ties to academic hosts including Vienna University of Technology and national bodies such as the Austrian Academy of Sciences. Due to global public health responses influenced by organizations like the World Health Organization and national authorities in Austria, the conference pivoted to an online format hosted on platforms used by events affiliated with Zoom Video Communications, YouTube, and virtual conference services employed by ACM. The program committee included area chairs and reviewers with affiliations spanning Princeton University, ETH Zurich, University College London, Imperial College London, and Peking University.

Program and Keynote Speakers

The keynote roster highlighted influential figures from academia and industry, including speakers connected to DeepMind teams led by researchers related to work appearing alongside names such as Demis Hassabis and contributors from Google DeepMind projects, as well as speakers with links to OpenAI leadership and research groups associated with Ilya Sutskever and Dario Amodei. Other presenters had affiliations with Facebook AI Research leadership and research labs tied to Yann LeCun and Meta Platforms, Inc. adjunct researchers. The invited talks echoed themes explored by scholars connected to the Turing Award community, and practitioners from companies including Apple Inc., Intel Corporation, and Microsoft presented perspectives bridging industry deployments and academic research.

Accepted Papers and Highlights

Accepted papers spanned topics with lineage tracing to work from labs like Google Research's Brain team, DeepMind's RL groups, and university groups at MIT, Stanford University, and University of Cambridge. Highlights included contributions on self-supervised learning with antecedents in projects related to BERT and architectures linked to Transformer (machine learning model), advances in graph neural networks building on research from groups such as Yoshua Bengio's collaborators, and optimization techniques extending the legacy of stochastic gradient research tied to scholars from Courant Institute and University of California, Berkeley. Papers addressed robustness and fairness with conceptual ties to discussions pursued at AAAI and IJCAI, and work on scalable training referenced systems developed in concert with NVIDIA and cloud platforms like Google Cloud Platform.

Workshops and Tutorials

The satellite program included workshops and tutorials reflecting specialized communities associated with venues like NeurIPS workshops and educational efforts similar to programs run by The Alan Turing Institute and Simons Institute for the Theory of Computing. Sessions covered reinforcement learning methods related to prior conferences hosted by ICLR, probabilistic modeling traditions stemming from UAI, privacy-preserving learning influenced by initiatives at OpenMined, and applications in healthcare and biology connecting to collaborations with Broad Institute and Wellcome Trust. Tutorials were delivered by instructors with affiliations including Columbia University, Yale University, Johns Hopkins University, and industry educators from AWS and Facebook.

Awards and Notable Contributions

ICML 2020 presented awards aligned with the conference's history of recognizing impactful research, building on a lineage of honors comparable to the Turing Award-associated recognition in the field. Best paper and outstanding paper awards highlighted teams from Stanford University, University of Toronto, and ETH Zurich. Notable methodological contributions influenced subsequent work at venues such as NeurIPS 2020 and ICLR 2021, and were adopted in engineering projects at Google, Microsoft, and startups incubated in ecosystems like Y Combinator and Sequoia Capital-backed firms. The conference furthered collaborations across academic groups connected to Max Planck Institute for Intelligent Systems, Blavatnik School of Government adjacent research, and industrial research labs such as DeepMind and Facebook AI Research.

Category:Machine learning conferences