OpenAI's Jeff Clune shares a new ML algorithm, ANML, which can learn 600 sequential tasks with minimal catastrophic forgetting, a long-standing challenge of AI
People learn differently from neural networks. If a human comes back to a sport after years away, they might be rusty …
Learning to Continuously Learn. ANML meta-learns to reduce catastrophic forgetting, and can learn at least 600 tasks (Omniglot classes) sequentially and performs well on most afterwards. arXiv https://arxiv.org/... NeurIPS talk https://slideslive.com/... Led by Shawn Beaulieu 1/2…
@epro
Emil Protalinski on x
“I think that these sorts of advances will be used in almost every situation where we use AI. It will just make AI better.” — @OpenAI's Jeff Clune on ANML. https://venturebeat.com/...
@usmansattarmd
Usman Sattar on x
@OpenAI 's @jeffclune on #deeplearning ‘s Achilles’ heel and a faster path to #AGI “This week, they collectively shared ANML (a neuromodulated meta-learning algorithm), which is able to learn 600 sequential tasks with minimal catastrophic forgetting” https://venturebeat.com/...