Name
Abstract | ||
|---|---|---|
The "Interested Reader" Model We present a simple, easily implemented spectral learning algorithm which applies equally whether we have no supervisory information, pairwise link constraints, or labeled examples. In the unsuper vised case, it performs consistently with other spec tral clustering algorithms. In the supervised case, our approach achieves high accuracy on the cate gorization of thousands of documents given only a few dozen labeled training documents for the 20 Newsgroups data set. Furthermore, its classifica tion accuracy increases with the addition of unla beled documents, demonstrating effective use of unlabeled data. By using normalized affinity ma trices which are both symmetric and stochastic, we also obtain both a probabilistic interpretation of our method and certain guarantees of performance. |
Year | Venue | Keywords |
|---|---|---|
2003 | IJCAI | classification accuracy increase,spectral clustering algorithm,unlabeled document,certain guarantee,unsupervised case,Newsgroups data,unlabeled data,supervised case,spectral learning,effective use,high accuracy |
DocType | Citations | PageRank |
Conference | 90 | 6.89 |
References | Authors | |
7 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Sepandar D. Kamvar | 1 | 2710 | 197.74 |
| Dan Klein | 2 | 8083 | 495.21 |
| Christopher D. Manning | 3 | 22579 | 1126.22 |