Loading...
Please wait, while we are loading the content...
Similar Documents
GloBoost Combinaisons de moindres généralisés : Produire des moindres généralisés et les combiner au sein d'un vote pondéré
| Content Provider | Semantic Scholar |
|---|---|
| Author | Torre, Fabien |
| Copyright Year | 2005 |
| Abstract | The primary goal of this paper is to propose a new learner for boosting algorithms, namely least general generalization. First experiments conducted on benchmarks show that ADA BOOST boosting least general generalization obtains smaller error than reference systems. The computation time needed by these experiments leads us to define a method that could be easily distributed. This method, called GloBoost, is able to produce hypotheses independently of one another and then give a weight to each produced hypothesis. New experiments are then conducted and show low error rates for both ADABOOST and GLOBOOST. Moreover, GLOBOOST has the advantage to be naturally distributable to different computers. |
| Starting Page | 769 |
| Ending Page | 797 |
| Page Count | 29 |
| File Format | PDF HTM / HTML |
| Volume Number | 19 |
| Alternate Webpage(s) | http://www.grappa.univ-lille3.fr/~torre/Recherche/Articles/2004/GloBoost-RIA.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
