Target-driven Visual Navigation in Indoor Scenes using Deep Reinforcement Learning
In this video, we introduce a deep reinforcement learning model for visual navigation. We also demonstrate AI2-THOR framework, an environment with high-quality 3D scenes and physics engine that enables interaction with objects.
Видео Target-driven Visual Navigation in Indoor Scenes using Deep Reinforcement Learning канала Allen Institute for AI
Видео Target-driven Visual Navigation in Indoor Scenes using Deep Reinforcement Learning канала Allen Institute for AI
Показать
Комментарии отсутствуют
Информация о видео
Другие видео канала
Language AI for RNA Virus and RNA VaccineOpenWebMath: An Open Dataset of High-Quality Mathematical Web TextOn Parameter Efficiency of Neural Language ModelsStudying Large Language Model Generalization with Influence FunctionsModular Language ModelsSkill it! A Data-Driven Skills Framework for Understanding and Training Language ModelsFrom Compression to Convection: A Latent Variable PerspectiveGenerative AI & CopyrightDo language models have coherent mental models of everyday things?Imaginative Vision Language ModelsStructure Modeling in Language ModelsWhen Not to Trust Language Models: Investigating Effectiveness of Parametric&Non-Parametric MemoriesEnhancing the Reliability and Continual Improvement of Neural Dialogue SystemsToward Intelligent Writing Support Beyond Completing SentencesTowards robust long-form text generation systemsEvaluating ethical and social risks from large models"Open" AI: considering the ethical upsides and downsides of "Open" AI developmentThe BigScience WorkshopMoving Forward by Moving Backward: Embedding Action Impact over Action Semantics | AI2Recycling finetuned models to pretrain: on loss spaces, fusing and evolving pretrainingRobot learning and perception for contact-rich manipulation