How Uber Uses Reinforcement Learning

This article discusses how Uber utilizes reinforcement learning techniques to enhance the efficiency of its marketplace by improving the balance between drivers and demand.

The article introduces Fiber, a scalable distributed computing framework designed to enhance the efficiency and flexibility of reinforcement learning (RL) and population-based methods.

The article discusses Enhanced POET, an advanced open-ended reinforcement learning algorithm that autonomously generates diverse learning challenges and solutions.

The article discusses a novel approach to training conversational agents using reinforcement learning, focusing on their ability to communicate solely through self-generated language.

The Second Uber Science Symposium showcased advancements in programming systems and tools, featuring talks from leading researchers and practitioners from institutions like MIT and Berkeley, as wel...

The Second Uber Science Symposium focused on advances in behavioral science, featuring presentations from leading researchers in the field.

The First Uber Science Symposium brought together experts from various fields to discuss advancements in reinforcement learning (RL), natural language processing (NLP), conversational AI, and deep ...

The article discusses the creation of an Atari model zoo aimed at enhancing the understanding of deep reinforcement learning (deep RL).

The article discusses the Atari Model Zoo, a framework designed to analyze, visualize, and compare deep reinforcement learning agents.

The article discusses Deep Curiosity Search (DeepCS), a novel approach in deep reinforcement learning (RL) that emphasizes intra-life exploration to enhance agent performance in challenging environ...

The article discusses the limitations of action-dependent baselines in reinforcement learning, specifically how they do not reduce variance compared to state-dependent baselines.

The article discusses the effectiveness of genetic algorithms (GAs) as a competitive alternative to traditional gradient-based methods for training deep neural networks (DNNs) in reinforcement lear...

The article discusses the Interpolated Policy Gradient method, which combines on-policy and off-policy gradient estimation techniques in deep reinforcement learning.

The article discusses advancements in image captioning techniques, highlighting the limitations of existing methods and proposing a new framework based on Conditional Generative Adversarial Network...