What once started as a small conference with a few hundred people interested in neural information processing systems has over the years turned into one of the largest artificial intelligence conferences in the world. This year, over 13,000 people attended the 33rd conference on Neural Information Processing Systems (NeurIPS) in Vancouver, British Columbia. Over the course of the week, ML@GT presented 30 papers through poster sessions, oral and spotlight presentations, and workshops.
While it’s impossible to cover every minute of the conference, here are a few of ML@GT’s favorite moments from this year’s conference.
Mohammad Reza Keshtkaran (Reza) walked crowds of people through his paper, “Enabling Hyperparameter Optimization in Sequential Autoencoders for Spiking Neural Data” This work was done in collaboration with his advisor, Chethan Pandarinath.
While it rained most of the week, the week started off with a bit of sunshine and blue skies.
We celebrated a great week with many of our students, faculty, alumni, and friends of the center at the ML@GT NeurIPS Lunch. Thanks to TAPshack Coal Harbour for its excellent food and atmosphere!
Saurabh Sawlani answered questions and discussed “Faster Width-Dependent Algorithm for Mixed Packing and Covering LPS” during Wednesday’s poster session. Sawlani also presented this work Wednesday morning during an oral presentation.
Minshuo Chen described his work on “Efficient Approximation of Deep ReLU Networks for Functions on Low Dimensional Manifolds” during Tuesday night’s poster session. For more information on this work, read his blog post.
Georgia Tech alumna Ashley Edwards stopped by Ayush Shrivastava’s poster on “Chasing Ghosts: Instruction Following as Bayesian State Tracking.” Learn more about this work on teaching robots to move here.
Eva Dyer and Chris Rozell supported their student and first author John Lee, during the poster session. The trio answered questions on their paper, “Hierarchical Optimal Transport for Multimodal Distribution Alignment” to a consistently packed crowd.
Poster sessions are a great opportunity to have 1×1 conversations with authors, support colleagues work, and more. Weiyang Liu reconnected with old friends while also discussing his latest work on “Bayesian Meta-network Architecture Learning” during Thursday’s poster session.
Georgia Tech alumna Ashley Edwards reconnected with her former advisor, Charles Isbell, his wife and GTRI software engineering and analytics chief Sheila, and NeurIPS communications co-chair Michael Littman. (Photo courtesy of Ashley Edwards)
Despite the rain, we could not complain about the views from the Vancouver Convention Center.
For more information on Georgia Tech’s papers at NeurIPS 2019, please click here.
We’ll see you all again in Vancouver next year!