NeurIPS 2014

NeurIPS 2014
Name	Neural Information Processing Systems 2014
Other names	NIPS 2014
Venue	Convention Center
Location	Montreal, Quebec
Country	Canada
Date	December 8–13, 2014
Organizers	Neural Information Processing Systems Foundation
Previous	Neural Information Processing Systems 2013
Next	Neural Information Processing Systems 2015

NeurIPS 2014 Neural Information Processing Systems 2014 was the 28th annual meeting bringing together researchers from Geoffrey Hinton's era of deep learning resurgence and communities around Yoshua Bengio, Yann LeCun, Andrew Ng, Michael Jordan (computer scientist), and Judea Pearl. The conference in Montreal showcased advances linking work from groups at University of Toronto, University of Montreal, Stanford University, Massachusetts Institute of Technology, and industry labs such as Google DeepMind, Facebook AI Research, Microsoft Research, and IBM Research. Researchers who had published at venues like ICML, CVPR, ACL, ICLR, and AAAI converged to discuss themes including representation learning, graphical models, optimization, and reinforcement learning.

Overview

NeurIPS 2014 continued traditions established by the Neural Information Processing Systems Foundation and followed formats similar to earlier meetings like NIPS 2013 while setting a stage for subsequent gatherings such as NIPS 2015. The program drew participants from academic institutions including Carnegie Mellon University, Princeton University, Columbia University, Harvard University, University of California, Berkeley, University of Washington, University of Oxford, University of Cambridge, ETH Zurich, EPFL, and Tsinghua University. Industry presence included teams from Google, Facebook, Microsoft, Amazon, Apple, DeepMind, Intel, NVIDIA, and Adobe. Attendees included scholars with affiliations to research centers like Vector Institute, CIFAR, Allen Institute for Artificial Intelligence, and Toyota Research Institute.

Conference Program and Keynotes

Keynote lectures featured prominent figures associated with breakthroughs in probabilistic modeling and deep architectures, echoing prior keynote histories of Yoshua Bengio, Yann LeCun, and Geoffrey Hinton. Invited talks and panels drew connections to work by researchers from Stanford University, University of Toronto, University of Montreal, and companies such as Google Research and Facebook AI Research. Sessions were organized into tracks reflecting themes familiar from ICML and CVPR programs: optimization techniques, generative models, unsupervised learning, and reinforcement learning related to work by Richard Sutton, Peter Dayan, and David Silver. Panels and plenaries included contributors with careers linked to institutions like MIT, Berkeley AI Research (BAIR), Oxford Machine Learning Research Group, and Max Planck Institute for Intelligent Systems.

Accepted Papers and Notable Contributions

Accepted papers spanned topics with lineage to work by Judea Pearl on causality, David Rumelhart on connectionism, and researchers connected to CIFAR programs. Important contributions included advances in deep convolutional models related to Yann LeCun's convolutional networks, variational methods extending ideas from Diederik Kingma and Max Welling, and structure-learning papers building on graphical-model traditions from Steffen Lauritzen and Michael Jordan (computer scientist). Reinforcement learning papers referenced foundations by Richard Sutton and applied methods that informed later projects at DeepMind and OpenAI. Optimization work connected to results by Leon Bottou and John Duchi on stochastic methods. Other accepted work related to dimensionality reduction traditions from Laurens van der Maaten and Geoffrey Hinton's earlier work on autoencoders.

Workshops, Tutorials, and Competitions

The workshop program included community-driven events influenced by groups such as CIFAR and organizers from University of Toronto, University of Montreal, NYU Machine Learning Group, and ETH Zurich. Tutorials covered practical techniques and theoretical frameworks similar to material presented at ICLR and ICML courses, often led by researchers affiliated with Google and Microsoft Research. Competitions and shared tasks reflected evaluation practices seen in ImageNet and Kaggle challenges, and workshops fostered collaborations with initiatives like OpenAI, Allen Institute for AI, Stanford AI Lab, and the Berkeley Artificial Intelligence Research community.

Organization and Attendance

The meeting was organized under the auspices of the Neural Information Processing Systems Foundation with program committees drawn from universities such as University of Toronto, Stanford University, MIT, UC Berkeley, University of Montreal, and industry labs including Google DeepMind, Facebook AI Research, Microsoft Research, and IBM Research. Attendance included graduate students, postdoctoral researchers, faculty, and engineers from institutions like Carnegie Mellon University, Columbia University, Harvard University, University of Oxford, University of Cambridge, EPFL, ETH Zurich, Tsinghua University, Peking University, and research groups at Yahoo Research and Intel Labs. Conference logistics engaged Montreal-based organizations and venues aligned with the city's history of hosting scientific meetings.

Impact and Legacy

NeurIPS 2014 is remembered for consolidating trajectories in deep learning promoted by Geoffrey Hinton, Yoshua Bengio, and Yann LeCun and for work that influenced later milestones at DeepMind (including projects led by Demis Hassabis), OpenAI initiatives, and industry research at Google Brain and Facebook AI Research. Research presented at the conference fed into subsequent developments at venues such as ICLR 2015 and ICML 2015 and influenced funding priorities at organizations like CIFAR and national research agencies. The event strengthened collaborations among labs at University of Toronto, University of Montreal, Stanford University, MIT, and Berkeley, and seeded projects that later appeared in applied settings across Google, Facebook, Microsoft, Amazon, and Apple.

Category:Neural Information Processing Systems conferences