NVIDIA at NeurIPS 2021

At the forefront of AI innovation, NVIDIA continues to push the boundaries of technology in machine learning, self-driving cars, robotics, graphics, and more. NVIDIA researchers will present 20 papers at NeurIPS from December 6 to December 14, 2021. Join us to see the latest advancements in research.

Latest in AI and Graphics

Virtual Research Presentations

Our accepted papers feature a range of groundbreaking research. From Alias-Free GANs creating photorealistic images to sematic segmentation with transformers, explore the exceptional work we’re bringing to the NeurIPS community.

Alias-Free Generative Adversarial Networks

Tero Karras, Miika Aittala, Samuli Laine, Erik Härkönen, Janne Hellsten, Jaakko Lehtinen, Timo Aila | Paper

EditGAN: High-Precision Semantic Image Editing

Huan Ling*, Karsten Kreis*, Daiqing Li, Seung Wook Kim, Antonio Torralba, Sanja Fidler | Paper

Score-Based Generative Modeling in Latent Space

Arash Vahdat*, Karsten Kreis*, Jan Kautz | Paper

Learning 3D Dense Correspondence via a Canonical Point Autoencoder

An-Chieh Cheng, Xueting Li, Min Sun, Ming-Hsuan Yang, Sifei Liu | Paper

Don’t Generate Me: Training Differentially Private Generative Models with Sinkhorn Divergence

Tianshi Cao, Alex Bie, Arash Vahdat, Sanja Fidler, Karsten Kreis | Paper

Controllable and Compositional Generation with Latent-Space, Energy-Based Models

Weili Nie, Arash Vahdat, Anima Anandkumar | Paper

Dynamic Bottleneck for Robust, Self-Supervised Exploration

Chenjia Bai, Lingxiao Wang, Lei Han, Animesh Garg, Jianye Hao, Peng Liu, Zhaoran Wang | Paper

Neural Hybrid Automata: Learning Dynamics with Multiple Modes and Stochastic Transitions

Michael Poli, Stefano Massaroli, Luca Scimeca, Seong Joon Oh, Sanghyuk Chun, Atsushi Yamashita, Hajime Asama, Jinkyoo Park, Animesh Garg | Paper

Drop-DTW: Aligning a Common Signal Between Sequences While Dropping Outliers

Nikita Dvornik, Isma Hadji, Konstantinos G. Derpanis, Animesh Garg, Allan D. Jepson | Paper

Deep Marching Tetrahedra: A Hybrid Representation for High-Resolution, 3D-Shape Synthesis

Tianchang Shen, Jun Gao, Kangxue Yin, Ming-Yu Liu, Sanja Fidler | Paper

ATISS: Autoregressive Transformers for Indoor-Scene Synthesis

Despoina Paschalidou, Amlan Kar, Maria Shugrina, Karsten Kreis, Andreas Geiger, Sanja Fidler | Paper

Toward Optimal Strategies for Training Self-Driving Perception Models in Simulation

David Acuna, Jonah Philion, Sanja Fidler | Paper

Ultrahyperbolic Neural Networks

Marc Law | Paper

SegFormer: Simple and Efficient Design for Semantic Segmentation with Transformers

Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M. Alvarez, Ping Luo | Paper

Adversarially Robust 3D-Point-Cloud Recognition Using Self-Supervisions

Jiachen Sun, Yulong Cao, Christopher Choy, Zhiding Yu, Anima Anandkumar, Z. Morley Mao, Chaowei Xiao | Paper

AugMax: Adversarial Composition of Random Augmentations for Robust Training

Haotao Wang, Chaowei Xiao, Jean Kossaifi, Zhiding Yu, Anima Anandkumar, Zhangyang Wang | Paper

Long-Short Transformer: Efficient Transformers for Language and Vision

Chen Zhu, Wei Ping, Chaowei Xiao, Mohammad Shoeybi, Tom Goldstein, Anima Anandkumar, Bryan Catanzaro | Paper

Personalized Federated Learning with Gaussian Processes

Idan Achituve, Aviv Shamsian, Aviv Navon, Gal Chechik, Ethan Fetaya | Paper

Improve Agents Without Retraining: Parallel Tree Search with Off-Policy Correction

Assaf Hallak, Gal Dalal, Steven Dalton, Iuri Frosio, Shie Mannor, Gal Chechik | Paper

DIB-R++: Learning to Predict Lighting and Material with a Hybrid Differentiable Renderer

Wenzheng Chen, Joey Litalien, Jun Gao, Zian Wang, Clement Fuji Tsang, Sameh Khamis, Or Litany, Sanja Fidler | Paper

Featured Demos



The software is one of the first to combine multiple modalities—text, semantic segmentation, sketch, and style—within a single GAN framework. This makes it faster and easier to turn an artist’s vision into a high-quality, AI-generated image.

NVIDIA Omniverse for Project Maxine

NVIDIA Omniverse for Project Maxine

To enable better communication and understanding, Project Maxine integrates NVIDIA Riva’s real-time translation and text-to-speech with photo animation “live portrait” and eye contact in real time.

Reconstructing 3D Environments from 2D Images

Reconstructing 3D Environments from 2D Images

NVIDIA Metropolis leverages streaming video to detect, track, infer 3D pose, and reconstruct full 3D scenes.

Robotics Advances with Visual Reasoning and Conversational AI

Robotics Advances with Visual Reasoning and Conversational AI

In this demo, Gemini, a robotic system that integrates both dialogue and visual reasoning, retrieves objects and hands them to a person.

GANverse3D: Knight Rider’s KITT Re-created with AI by NVIDIA

Bringing Knight Rider’s KITT to Life with AI

NVIDIA GANverse3D is a powerful breakthrough for 2D-to-3D models. Now, it's possible to create a 3D, animatable model with a single 2D picture.​

Creating Real-Time Digital Avatars

Creating Real-Time Digital Avatars

Learn how this AI technology, which won Best in Show at SIGGRAPH 2021, can be used for video conferencing, storytelling, virtual assistants, and more.​



Collections from the NVIDIA NGC catalog are curated in easy-to-use packages that make it easy to discover compatible framework containers, models, Jupyter Notebooks, and other resources to get you started faster with your AI use cases.

NVIDIA Maxine​

NVIDIA Maxine​

NVIDIA Maxine is a GPU-accelerated SDK with state-of-the-art AI features for building virtual collaboration and content creation applications, such as video conferencing and live streaming.

Trainings & Resources

NVIDIA Developer Program

NVIDIA Developer Program

Get the advanced tools and training you need to successfully build applications on all NVIDIA technology platforms.

Unlock Your Startup’s Potential

Unlock Your Startup’s Potential

NVIDIA Inception nurtures cutting-edge technology startups that are revolutionizing industries. The program offers go-to-market support, expertise, and technology—all tailored to a business’s evolution, no matter what size or funding stage they’re at. 

NVIDIA Applied Research Accelerator Program

NVIDIA Applied Research Accelerator Program

The NVIDIA Applied Research Accelerator Program supports research projects that can make a real-world impact through deployment in GPU-accelerated applications adopted by commercial and government organizations.

Accelerate 3D Deep Learning Research with New Kaolin Features

Accelerate 3D Deep Learning Research with New Kaolin Features

Kaolin is launching new features to accelerate 3D deep learning research. Updates to the NVIDIA Omniverse Kaolin app will bring robust visualization of massive point clouds. Updates to the Kaolin library will include support for tetrahedral meshes, rays management functionality, and a strong speedup to DIB-R.

Generating Synthetic Data for Training Perception Models

Generating Synthetic Data for Training Perception Models

Training perception models requires extremely large and complex datasets. Assembly can be costly, time-consuming, dangerous and even impossible for certain cases. Developers can take easy to understand parameters and synthetically generate data for machine learning through the use of Omniverse Replicator for Isaac Sim.


When you work at NVIDIA, you have the opportunity to solve some of the world’s hardest problems and discover never-before-seen ways to improve the quality of life for people everywhere—from healthcare to robots. self-driving cars to blockbuster movies, and a growing list of new opportunities every single day.

Learn more about our career opportunities by exploring current job openings and university jobs.

Sign up to receive the latest news from NVIDIA.