An example of semi-supervised graphical model and alternating inference algorithm on a small-scale dataset
Run test.m to go over the process.
*************************************************************
The following packages are used for learning the tree model and graphical inference:
latentTree:
http://people.csail.mit.edu/myungjin/latentTree.html
UGM:
http://www.di.ens.fr/~mschmidt/Software/UGM.html
The code has been tested in Windows Server 2008 R2. It should be compatible also for Linux and Mac.
Compile (mex) the above packages before running the code.
The source code for the above packages is included for the ease of the readers. All the terms and conditions of the original packages apply. If you want to get the either of the packages removed, please contact yuxinnan@ee.columbia.edu, and we will remove it immediately.
******************************************************
For structural learning as described in the paper, we recommend BMRM (http://users.cecs.anu.edu.au/~chteo/BMRM.html) for C/C++ users, and Andrea Vedaldi's Matlab wrapper (http://www.vlfeat.org/~vedaldi/code/svm-struct-matlab.html) with SVM struct (http://www.cs.cornell.edu/People/tj/svm_light/svm_struct.html) for Matlab users.
******************************************************
Please cite the following papers when using the data/code:
Felix X. Yu; Rongrong Ji; Ming-Hen Tsai; Guangnan Ye; Shih-Fu Chang.
Weak attributes for large-scale image retrieval, CVPR 2012
Please also consider to cite our technical report covering more experiments:
Felix X. Yu; Rongrong Ji; Ming-Hen Tsai; Guangnan Ye; Shih-Fu Chang.
Experiments of image retrieval using weak attributes, Technical Report # CUCS 005-12
*************************************************
Report bugs/questions to Felix X. Yu: yuxinnan@ee.columbia.edu
资源2 UGM: A Matlab toolbox for probabilistic undirected graphical models:http://www.cs.ubc.ca/~schmidtm/Software/UGM.html
这是UBC.ca的博士M. Schmidt做的MATLAB Toolbox, 内容上很不错
UGM is a set of Matlab functions implementing various tasks in probabilistic undirected graphical models of discrete data with pairwise (and unary) potentials. Specifically, it implements a variety of methods for the following four tasks:
Decoding: Computing the most likely configuration.
Inference: Computing the partition function and marginal
probabilities.
Sampling: Generating samples from the distribution.
Training: Fitting a model to a given dataset.
资源2 UGM还有关于MRF和CRF的资料,记得VALSE2015.8.13的Panel主题:
如何做研究之读博经验纵横谈(http://valseonline.org/article-75-1.html),牛津大学的郑帅博士(http://kylezheng.org/)也对CRF做过研究,并有CRF
as Deep learning's RNN的文章).
Documentation and Tutorial on Markov Random Fields and Conditional
Random Fields
资源3: http://people.csail.mit.edu/myungjin/latentTree.html
Learning Latent Tree Graphical Models
Massachusetts Institute of Technology