Findings: 可视皮层分层
1981
年的诺贝尔医学奖,颁发给了 David Hubel(出生于加拿大的美国神经生物学家)和TorstenWiesel,以及 Roger Sperry。前两位的主要贡献,是“发现了视觉系统的信息处理”:可视皮层是分级的:之所以做这个试验,目的是去证明一个猜测。位于后脑皮层的不同视觉神经元,与瞳孔所受刺激之间,存在某种对应关系。一旦瞳孔受到某一种刺激,后脑皮层的某一部分神经元就会活跃。经历了很多天反复的枯燥的试验,同时牺牲了若干只可怜的小猫,David Hubel 和Torsten Wiesel 发现了一种被称为“方向选择性细胞(Orientation Selective Cell)”的神经元细胞。当瞳孔发现了眼前的物体的边缘,而且这个边缘指向某个方向时,这种神经元细胞就会活跃。
这个发现激发了人们对于神经系统的进一步思考。神经-中枢-大脑的工作过程,或许是一个不断迭代、不断抽象的过程。
visual cortex
is constituted of a cell pattern, which appearsdesigned specifically for
vision.
Hubel and
Wiesel most famous studies were in the area of visualperception, with particular
emphasis on the nerve impulses mediating betweenthe retina and the brain. They
observed that various nerve cells wereresponsible for different types of visual
stimuli.【what’s the relationshipwith multilayer ??? 】
可以记忆形状、方向;研究者:
中国
A.
Jeff Hawkins
创办 - berkeleyredwood , numenta.org , Groksolutions.com
B.
Ideas
HTM is a
concept-understanding/memorise patternof the brain
C.
MainContent:
http://yukuan.blogspot.com/2006/06/on-intelligence.html
Bottom
up / Top down - as 生物神经网络基本原则 - from ever《数学之美番外篇:平凡而又神奇的贝叶斯方法》
分级抽象、时间流、序列、pattern概念
UCB
presentation - http://www.youtube.com/watch?v=TDzr0_fbnVk
D.
knowledgeinvolved
machine
learning , statistics , neural science ,neural computation -
http://redwood.berkeley.edu/wiki/VS298:_Slides
E.
some sources
Dileep
George - how the brain might work
On Intelligence
benifit a lot here -
http://blog.csdn.net/scyscyao/article/details/5863920 - ormy ever <分享几个Neural Computing和HTM的链接>
http://yukuan.blogspot.com/2006/06/on-intelligence.html
HTM学习笔记和疑问-具体请查看HTM学习笔记
Classify: that evil call up system
that such
brain just evolve to be so , with better adaptability
群虻行为很类似于大脑机制,单个虻的原则特别简单,数量多之后产生有效的生产模式,而群体是自然筛选出,可以持续生存的。
STDP ( Spike-Timing DependentPlasity ) : is a biological process that adjusts the strengthof connections
between neurons in the brain.
l
How the brain
works
l
artificial
minds - many aspect of my –interests - link
Sleep
; short term of study ; exercise– walking; sex
spaun大脑由250万个模拟神经元组成,它能执行8种不同类型的任务。这些任务的范围从描摹到计算,再到问题回答和流体推理(fluid reasoning)等,结果发表在Science上面。
1) Nengo
模拟神经系统的图形化开源软件包,需要超过台式机的计算能力。from canada uwaterloo university , is a project to simulate the
humanbrain in a supercomputer, using neural engineer , basic presentation ?
linearor vector, transformation, dymanic
2) Opensource
github ; python api or java toolkit
3) Spaun
Based on NENGO, a brainsimulation that have some features , give
out what does it see , remember and recognizewith eyes and hands
最初是mammal Brain Project , 之后是蓝脑计划,为模拟大脑,探索一些脑机制以及脑特性、模拟脑疾等
Opensource -
code : http://www.numenta.com/phpBB2/viewtopic.php?t=1285
初识时不会站立不会行走,经过各种尝试以及记忆,更新淘汰,直到可以直立行走,来源暂时没有找到
持续更新,欢迎指教、探讨、转载~
转载请注明出处
Notes on Brain Study,布布扣,bubuko.com
原文:http://www.cnblogs.com/hphp/p/3589822.html