Name
Abstract | ||
|---|---|---|
Mihalcea [1] discusses self-training and co-training in the context of word sense disambiguation and shows that parameter optimization on individual words was important to obtain good results. Using smoothed co-training of a naive Bayes classifier she obtains a 9.8% error reduction on Senseval-2 data with a fixed parameter setting. In this paper we test a semi-supervised learning algorithm with no parameters, namely tri-training [2]. We also test the random subspace method [3] for building committees out of stable learners. Both techniques lead to significant error reductions with different learning algorithms, but improvements do not accumulate. Our best error reduction is 7.4%, and our best absolute average over Senseval-2 data, though not directly comparable, is 12% higher than the results reported in Mihalcea [1]. |
Year | DOI | Venue |
|---|---|---|
2010 | 10.1007/978-3-642-14770-8_43 | IceTAL |
Keywords | Field | DocType |
word sense disambiguation,best absolute average,semi-supervised learning algorithm,senseval-2 data,error reduction,parameter optimization,significant error reduction,best error reduction,ensemble-based method,discusses self-training,different learning algorithm,fixed parameter setting,naive bayes,semi supervised learning | Naive Bayes classifier,Pattern recognition,Random subspace method,Computer science,Support vector machine,Co-training,Speech recognition,Artificial intelligence,Word-sense disambiguation | Conference |
Volume | ISSN | ISBN |
6233 | 0302-9743 | 3-642-14769-0 |
Citations | PageRank | References |
1 | 0.34 | 8 |
Authors | ||
2 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Anders Søgaard | 1 | 684 | 81.68 |
| Anders Johannsen | 2 | 150 | 12.12 |