The software for SVMs can be downloaded from www.joachims.org
Document Corpus Reuters 90 categories can be downloade from:
http://disi.unitn.it/moschitti/corpora.htm
1) Given the software for Support Vector Machines, implement at least 2 weighting schemes for Text Categorization task.
- Manually Optimize parameters on a held-out set
- Measure Precision, Recall, F1 on the test-set
- Experiment with n categories
-
Article 1 (show
as SVMs can be parameterized)
2) Given the software for Support Vector Machines implement at least 2 feature selectors.
- Manually Optimize parameters on a held-out set
- Measure Precision, Recall, F1 on the test-set
- Experiment with n categories
- Article 1 (show
as SVMs can be parameterized)
- Article 2
(feature selection)
3.a) Given the software for Support Vector Machines and data for some categories
- Optimize parameters using n-fold cross validation
- Measure Precision, Recall, F1 using n-fold cross validation
- Experiment
with n categories
- Articolo 1 (parameterization of SVMs with n-fold cross validation)
OR
4) Rocchio's classifier implementation
- Test on Reuters data
- Articolo 1 (shows how to implement Rocchio)
5) Naive Bayes' classifier implementation
- Test on Reuters data
- slides "basic concepts" or Andrea's slides
6) KNN's classifier implementation
- Test on Reuters data
- slides "text categorization" or Andrea's slides
7) Document Clustering
- Test on Reuters (or other) data
- slides "basic concepts" or Andrea's slides
- Articolo 1: (some techniques for basic clustering)
Given the following Software for Support Vector Machines:
- SVM-Light-TK1.2 available at http://disi.unint.it/moschitti/Tree-Kernel.htm
- SVM-Light-TK1.5 available in your home of lab account (or ask me).
- SVM-light 6.0 available at www.joachims.org
8) Given SVM-Light-TK1.2, implement string kernel (word sequenze kernel).
- Test on a small portion of Reuters
9) Given SVM-light 6.0 software, implement in it tree kernels.
- Experiment on Question Classification
10) Given SVM-light-TK-1.5 (built on top of svm-light 5.0), port it in SVM-light 6.0.
11) Implement feature selection based on SVMs (using SVM-Light-TK1.5)
12) Implement and Experiment with feature selection in Convolution Kernel Spaces (using SVM-Light-TK1.5)