Skip to main content

Uber AI

Uber AI is at the heart of AI-powered innovation and technologies at Uber. The research and advancements in artificial intelligence made by the team solve challenges across the whole of Uber.

Uber AI

Uber AI is at the heart of AI-powered innovation and technologies at Uber. The research and advancements in artificial intelligence made by the team solve challenges across the whole of Uber.

Uber AI Labs

Our world-class team at Uber AI Labs pursues fundamental research in machine learning and connects cutting-edge advances to the broader business. Machine learning is essential to our business, and we are therefore fully committed to the pursuit of fundamental advances and vigorous engagement with the broader machine learning community.

Computer vision

Cameras provide a unique insight into the complex world Uber operates in. Our mission is to drive product differentiation and business efficiencies at Uber using visual data. We work together with product teams to develop state of the art computer vision and machine learning solutions.

Conversational AI

Our mission is to make all Uber interactions as natural as talking to a friend! We deliver cutting-edge Conversational AI solutions to product teams across Uber, working with other AI teams and business units.

Sensing & Perception

We believe sensors tell a story. Our goal is to leverage sensors as a source of truth and develop algorithms to solve our users' top pain points. The team has improved features like crash detection, enhanced location accuracy innovations, and even sent phone mounts to drivers for safer driving.

Uber AI News

  • Modeling Censored Time-to-Event Data Using Pyro, an Open Source Probabilistic Programming Language

    Uber scientists demonstrate how to train time-to-event models from censored data using the Pyro probabilistic programming language.

  • First Uber Science Symposium: Discussing the Next Generation of RL, NLP, ConvAI, and DL

    Uber hosted the first-ever Uber Science Symposium, a full-day event of presentations and conversations with stellar speakers and attendees, focusing on reinforcement learning, natural language processing and conversational AI, deep learning and deep learning infrastructure.

  • Introducing Ludwig, a Code-Free Deep Learning Toolbox

    Uber AI introduces a toolbox to train deep learning models and use them for prediction that does not require any coding.

  • Creating a Zoo of Atari-Playing Agents to Catalyze the Understanding of Deep Reinforcement Learning

    To make research into understanding deep RL easier to conduct, by creating and open sourcing a repository of trained Atari Learning Environment agents along with tools to understand and analyze their behavior.

  • POET: Endlessly Generating Increasingly Complex and Diverse Learning Environments and their Solutions through the Paired Open-Ended Trailblazer

    Uber AI Labs introduces the Paired Open-Ended Trailblazer (POET), an algorithm that leverages open-endedness to push the bounds of machine learning.

  • How Uber Beacon Helps Improve Safety for Riders and Drivers

    Uber AI Labs discusses how beacon sensors help generate and validate vehicle movement data

  • Faster Neural Networks Straight from JPEG

    Ever thought of making CNNs for image classification smaller, faster, and more accurate, by simply hacking the JPEG codec? Also included are surprising insights about frequency space and color information as they relate to network architecture design.

  • How to Get a Better GAN (Almost) for Free: Introducing the Metropolis-Hastings GAN

    Generative Adversarial Networks (GANs) have achieved impressive feats in realistic image generation and image repair. At Uber, GANs have myriad potential applications, including strengthening our machine learning (ML) models against adversarial attacks, learning simulators for traffic, ride requests or demand patterns over time.

  • Montezuma’s Revenge Solved by Go-Explore, a New Algorithm for Hard-Exploration Problems (Sets Records on Pitfall, Too)

    In deep reinforcement learning (RL), solving the Atari games Montezuma’s Revenge and Pitfall has been a grand challenge. Today we introduce Go-Explore, a new family of algorithms capable of achieving scores over 2,000,000 on Montezuma’s Revenge and scoring over 400,000 on average!