CAREER: Biologically Inspired Learning Algorithms for Artificial Sensorimotor Systems

职业：人工感觉运动系统的生物启发学习算法

• 批准号: 0347132
• 负责人: Daniel Lee
• 金额: --
• 依托单位: University of Pennsylvania
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 2004
• 资助国家: 美国
• 起止时间: 2004-06-01 至 2010-05-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0347132&HistoricalAwards=false
• 关键词: CAREER; Biologically; Inspired; Learning; Algorithms

This project will investigate biologically-inspired methodologies and algorithms for several important problems in machine learning and AI: unsupervised feature learning, multi-modal integration of sensory data, and learning for real-time robotic systems with embedded sensors and actuators. The project will pursue both theoretical investigations of nonlinear properties of neurons and synapses and their role in processing of perceptual (visual, auditory, olfactory) stimuli, as well as the applications of such properties to the design of novel algorithms and architectures for processing of multi-modal perceptual stimuli. This research will integrate advances in the understanding of biological systems by building more advanced artificial sensorimotor systems. This project integrates research and an educational agenda that includes teaching and participation of students from the broader community in activities related to the overall goals of this research.

该项目将研究受生物启发的方法和算法，以解决机器学习和人工智能中的几个重要问题：无监督特征学习、传感数据的多模态集成以及嵌入式传感器和执行器的实时机器人系统的学习。该项目将对神经元和突触的非线性特性及其在处理感知（视觉、听觉、嗅觉）刺激中的作用进行理论研究，并将这些特性应用于设计用于处理多模态感知刺激的新算法和架构。这项研究将通过建立更先进的人工感觉运动系统来整合对生物系统的理解。该项目整合了研究和教育议程，包括教学和来自更广泛社区的学生参与与本研究总体目标相关的活动。