DeepMind enhances AI with new algorithm that incorporates ‘memory’ #artificialintelligence

DeepMind, known for its advancements in machine learning, initially taught its systems to play Atari games in 2014. However, the system could only learn to play one game at a time and couldn’t retain the information for multiple games. A recent collaboration between DeepMind and Imperial College London has led to the development of an algorithm that enables neural networks to learn, retain information, and apply it across different games. The research, published in the Proceedings of the National Academy of Sciences, showcases the AI’s ability to learn in sequences using supervised and reinforcement learning tests. This breakthrough allows the system to learn and play several games consecutively, marking a significant advancement in AI capabilities. The new algorithm represents a major step towards creating AI systems that can learn and adapt in a more versatile and efficient manner. For more details on this research, you can visit the original article on WIRED.com.

Source link

Source link: https://www.predictiveanalyticsworld.com/machinelearningtimes/deepminds-new-algorithm-adds-memory-to-ai/13433/