We are developing new deep learning methods to make sense of the world’s textual information. We focus on finding new ways to automatically extract relevant information from a large and diverse range of sources, and on building smarter systems for human-machine interactions.
We are creating new reinforcement learning algorithms that learn to make optimal decisions in complex environments. Our research allows us to deliver new and better solutions for a wide range of industries such as finance, manufacturing, and logistics.
Measuring uncertainties is essential for real-world applications of reinforcement learning. A computer that we depend on to make important decisions should account for the risks involved, and should also be able to give warning when it's not sure what the best decision is. Researchers at Uncharted Technologies have developed a new and practical method for measuring uncertainties, thus making a step closer to widespread real-world applications of reinforcement learning. Click here to read the research paper.
Read more ›
Reinforcement learning (RL) is a branch of AI in which an agent learns how to interact with its environment in order to maximize a reward signal. RL has successfully been used to reach superhuman performance at Go, chess, and a wide range of video games, and UncharTECH is on a mission to untap its potential for real world applications. We are pleased to share these learning notes that describe some key concepts in the field and some applications.
Read more ›
We shared the stage with Facebook and Google at the "Maths de l'IA" event at the Laboratoire de Mathématiques d'Orsay on the 13th of February 2019, attended by a large number of students and researchers in mathematics. Dr. William Clements, senior researcher at Unchartech, described some ways in which mathematics research can contribute to overcome the challenges faced by the reinforcement learning community.