Tag: Artificial Intelligence

Articles

Building the next generation of AI models to decipher human biology

The human genome, containing the entirety of our genetic blueprint, holds the keys to understand the intricate workings of our cells and bodies. How our cells respond to signals, build proteins, and fight off diseases, is determined by instructions encoded within our DNA. While scientists have made tremendous progress in deciphering this code, especially since… Read more »

Africa’s AI community gathers in Tunis for Deep Learning Indaba

Empowering African AI talent at in-person event for the first time post-covid InstaDeep played a key role as co-organiser of the Deep Learning Indaba, Africa’s biggest AI and machine learning workshop and conference this summer. Around 400 researchers, scientists, developers, academics and startup founders from across Africa attended the event at Sup’Com, the Higher School… Read more »

In this post, we introduce our first Meta-RL algorithm: MAML (Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks). With MAML, you can train agents that quickly adapt in almost any dense-reward environment. Let’s detail how it works.

Model Agnostic Meta-Learning made simple

(Part 2/4) In our introduction to meta-reinforcement learning, we presented the main concepts of meta-RL: Meta-Environments are associated with a distribution of distinct MDPs called tasks. The goal of Meta-RL is to learn to leverage prior experience to adapt quickly to new tasks. In Meta-RL, we learn an algorithm during a step called meta-training. At meta-testing, we apply this… Read more »

A simple introduction to Meta-Reinforcement Learning

A simple introduction to Meta-Reinforcement Learning

(Part 1/4) The recent developments in Reinforcement Learning (RL) have shown the incredible capacity of computers to outperform human performance in many environments such as Atari Games [1], Go, chess, shogi [2], Starcraft II [3]. This performance results from the development of Deep Learning and Reinforcement Learning methods like Deep Q-Networks (DQN) [4] and actor-critic methods [5, 6, 7, 8]. However, one essential advantage of… Read more »

InstaDeep at NeurIPS 2018

We have a strong team present at NeurIPS this week, with exciting things on the agenda. We are proud to announce that — for the first time — two of our research papers will be presented, a huge milestone for us. Additionally, three of our colleagues including our CEO, Karim, are presenting in the workshop Black in AI (BAI)…. Read more »