Artificial intelligence firm Deep Mind has started developing general artificial intelligence, which learns to accomplish different tasks without human input

Briefing

Artificial intelligence firm Deep Mind has started developing general artificial intelligence, which learns to accomplish different tasks without human input

August 22, 2018

Briefing
  • Artificial General Intelligence – DeepMind developing general artificial intelligence, called Impala, that can learn multiple new tasks without human intervention
  • Reinforcement Learning – Uses reinforcement learning, method in which algorithms are given rewards for certain actions and mastery of skills
  • A3C Algorithm – Reinforcement learning algorithm, known as A3C, supercharges learning process, as one neural network, the “critic,” evaluates “actor” neural network
  • V-Trace Algorithm – Correction algorithm called V-trace makes learning process more efficient by helping Impala to achieve positive transfer of learning between tasks

Accelerator
Artificial Intelligence

Sector
Information Technology

Organization
DeepMind Technologies Ltd.

Source

Original Publication Date
August 21, 2018

Leave a comment