我倒不是做人脑认知的。只是这学期有门课叫做 Uncertain Modeling of AI. 主要是讲BN的。老师是MIT本,博毕业的,所以
作业和project很多。
有几本书是值得一读的:
[1] (N) Neapolitan, R. E. Learning Bayesian Networks. Pearson Prentice
Hall, 2004
[2] (RN) Russell, S. and Norvig, P. Artificial Intelligence: A Modern
Approach, 2nd ed. Prentice Hall, 2003
[3](KN) Korb, K. B. and Nicholson, A. E. Bayesian Artificial
Intelligence. Chapman and Hall/CRC, 2004.
我觉得(KN)讲得比较详细。
会议和Murphy的主页:
You will also find valuable tutorials, tools, publications on
Bayesian networks and related technologies at the following
websites:
Conference in Uncertainty in Artificial Intelligence (UAI)
www.auai.org
American Association for Artificial Intelligence Conference (AAAI)
www.aaai.org
International Joint Conference on Artificial Intelligence (IJCAI)
www.ijcai.org
Neural Information Processing Systems Conference (NIPS)
www.nips.cc
Kevin Murphy’s tutorial on Graphical models and Bayesian networks
http://www.cs.ubc.ca/~murphyk/Bayes/bnintro.html
两个著名的研究BN的工具软件:
Genie 2.0 by Decision Systems Laboratory, University of Pittsburgh
http://genie.sis.pitt.edu/
An excellent, free probabilistic networks reasoning program
Netica 4.08 by Norsys Software Corp.
Free versions of the main application and the APIs are available
BN的无痛入门:
Charniak, Eugene “Bayesian Networks without Tears”, AI Magazine,
12(4), Winter 91, 50-63
Exactly Inference 是一定不能少了这伙计的: Pearl, BN的始祖之一
当然绝对不能少了Jordan和Bishop的材料(Approximate Inference):
[PRML] Pattern Recognition and Machine Learning 第10章Approximate
Inference 主要讲了Variational Method;
Jordan 派的Graphical Model这个自然就不用再说了。
On 9月1日, 下午4时01分, pongba <
pon...@gmail.com> wrote:
> 在人脑的认知方面,原则上总结得已经挺不错了,接下来就是怎样用量化的数学模型来描述我们的大脑了,因为如果不能用量化数学模型描述出来,又怎么能编写成程序呢?后者,是属于认知科学和计算机科学交界处的事情。就我的知识,目前有两派理论比较火,一是神经网络,这个是直接从人脑处理信息的物理基础切入。二是概率推理模型,这个是从抽象层面切入——一定要神经网络作为基础才能完成人脑所能完成的工作吗?
>
> 由于这个领域我目前的认识尚浅,言多必属扯淡,所以就说这些,神经网络我上次贴过一些
> reference<
https://groups.google.com/group/pongba/browse_thread/thread/93ac6f711...>,这次贴一些关于概率推理的:
> Kemp <
http://www.psy.cmu.edu/%7Eckemp/> 的主页,注意这几篇 paper:
>
> [1] Kemp 写的 Bayesian Model of
> Cognition<
http://www.psy.cmu.edu/%7Eckemp/papers/bayeschapter.pdf>,里面对贝叶斯推理作了非常漂亮和系统的介绍。(注:paper
> 里提到的 Bayesian Occam's Razor 理论在 David
> J.C. MacKay <
http://www.inference.phy.cam.ac.uk/mackay/> 的牛书 《Information
> Theory: Inference and Learning
> Algorithms》(作者开放电子版<
http://www.inference.phy.cam.ac.uk/mackay/itila/>)里面的第28章:"模型比较与奥卡姆剃刀"
> 中有精彩的阐述)
> [2] Tenenbaum 写的两篇综述性质的文章:1. Probabilistic models of cognition: Conceptual
> foundations<
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6VH9-4K8SC6...>2.
> Probabilistic
> models of cognition: where
> next?<
http://www.sciencedirect.com/science?_ob=ArticleURL&_udi=B6VH9-4K6CR0...>牛人写的综述是极有价值的,可以为一个领域指明方向。
>
> 最后要大力感谢 Shenli 同学上次分享的信息<
https://groups.google.com/group/pongba/browse_thread/thread/dacd6f1ae...>
> 。