← 返回论文列表 📄 下载原文 PDF  ISSCC 2008 · 1.4
ISSCC 2008Session 1 · PLENARYPlenary

Why Can’t A Computer Be More Like A Brain? Or What To Do With All Those Transistors? Jeff Hawkins

⚡ 本页包含 AI 生成的分析内容,仅供参考

📋 论文概要

本演讲探讨了传统计算机在人工智能任务上进展缓慢的原因,提出应借鉴大脑新皮质的结构和工作原理,利用大量晶体管构建基于记忆预测和层次化模式识别的类脑计算架构,以突破现有AI瓶颈。

💡 主要创新点

重要性
发表年份
ISSCC 2008

🏷 关键词

类脑计算新皮质记忆预测人工智能稀疏表示

📄 原文摘要

Why is it so difficult for computers to perform tasks that humans do quickly and easily? For over fifty years, we have tried to program computers to recognize images, understand language, control robots, and learn on their own. Although we have had some limited successes, what is more remarkable is how little progress we have made. Large increases in memory capacity and CPU performance have not resulted in corresponding improvements in performance on AI tasks. Machine intelligence is in need of new approaches. In humans, the neocortex is responsible for most high-level perception, language, and motor control. The anatomy and physiology of this large sheet of cells have been studied in sufficient detail to constrain how information must be stored and retrieved in neocortical tissue. The neocortex, indeed the entire nervous system, has a hierarchical organization. Knowledge about the world, whether

👥 作者与机构

Founder, Numenta, Menlo Park, CA, 1. Overview

分类:Plenary · 年份:ISSCC 2008