Coevolution of Agents and Environments (POET)
This video explains one of the most interesting ideas in Artificial Intelligence! The POET algorithm is a framework to simultaneously optimize agents while evolving the complexity of the environments they interact with. This algorithm has shown remarkable results in the Bipedal Walker task mapping 24 input variables (LIDAR sensors + Internal state variables) to 4 outputs controlling the walker's movements! Thanks for watching! Please Subscribe!
Paper Link: https://arxiv.org/pdf/1901.01753.pdf
Видео Coevolution of Agents and Environments (POET) канала Connor Shorten
Paper Link: https://arxiv.org/pdf/1901.01753.pdf
Видео Coevolution of Agents and Environments (POET) канала Connor Shorten
Показать
Комментарии отсутствуют
Информация о видео
Другие видео канала
Collaboration of ExpertsAI Weekly Update #12 - November 18th, 2019Healthsea from Spacy on HuggingFace SpacesDL Podcast #3 | Yannic Kilcher | Population-Based SearchDeep Learning Podcast #2 | Edward Peake | Deep Learning in Medical ImagingLong-Short TransformerKnowledge Distillation - Keras Code ExamplesApproximate Nearest Neighbor Benchmarks - Weaviate Podcast RecapBinary Passage Retrieval in Weaviate (32x Memory Savings)Beyond Goldfish Memory!Generating Wikipedia by Summarizing Long SequencesAI Weekly Update 2.0Weight Agnostic Neural Networks Explained!Efficient Transfer Learning with Null PromptsGoogle Research at ICCVDeep Learning for Podcast Content Search (Summary of Interview with Alex Canan at Zencastr)Full Stack Neural SearchThe Evolved TransformerEvolving Normalization-Activation LayersCoDA: Contrast-Enhancing and Diversity-Promoting Data Augmentation for NLUHow recent papers from OpenAI may come together