Uber AI

Artificial intelligence powers many of the technologies and services underpinning Uber’s platform, allowing engineering and data science teams to make informed decisions that help improve user experiences for products across our lines of business.

At Uber, many of the hard problems we work on can benefit from machine learning, such as improving safety, improving ETAs, recommending food items, and finding the best match between riders and drivers.

Plug and Play Language Model, introduced by Uber AI Labs, gives NLP practitioners the flexibility to plug in one or more simple attribute models into a large, unconditional language model.

Uber is presenting 11 papers at the NeurIPS 2019 conference in Vancouver, Canada, as well as sponsoring workshops including Women in Machine Learning (WiML) and Black in AI.

Uber AI's graph neural netowrk based method is used used for improving the quality of dish and restaurant recommendations in Uber Eats. The articledetails the algorithm, the experiments and the pipeline, and also shows examples of how they all work together in production to improve the user experience.Marketplace.

To facilitate the best end-to-end experience possible for users, Uber is committed to making customer support easier and more accessible. Working toward this goal, Uber’s Customer Obsession team leverages five different customer-agent communication channels powered by an in-house platform that integrates customer support ticket context for easy issue resolution.

Uber uses convolutional neural networks in many domains that could potentially involve coordinate transforms, from designing self-driving vehicles to automating street sign detection to build maps and maximizing the efficiency of spatial movements in the Uber Marketplace.

Uber launches Beacon V2, a piece of hardware for Uber drivers, that implements on device sensor fusion using signals from GNSS, IMU, and barometer to improve location accuracy.

Hypothesis-gufunc is an extension to hypothesis to aid in unit testing broadcasting and matrix input numpy/torch functions.

We develop a new measure called Loss Change Allocation (LCA) to better understand the training process. LCA measures how much each parameter helps decrease (or increase) the loss at every iteration. It is able to reveal broader patterns such as how training is very noisy, some layers hurt, and micro-learning is synchronized across layers.