Name
Abstract | ||
|---|---|---|
•We give a new proof for the finite termination of general decomposition methods for SVMs under a mild condition.•We improve the result of [16] in the sense that the relaxed KKT condition employed in [16] reduces to the commonly used one.•Our new convergence result can be applied to a wide class of decomposition algorithms, such as SMO and SVMlight algorithms. |
Year | DOI | Venue |
|---|---|---|
2018 | 10.1016/j.neucom.2018.08.030 | Neurocomputing |
Keywords | Field | DocType |
Support vector machines,Decomposition methods,Convergence,Quadratic programming,Finite termination | Convergence (routing),Applied mathematics,Finite set,Working set,Pattern recognition,Support vector machine,Hessian matrix,Artificial intelligence,Quadratic programming,Karush–Kuhn–Tucker conditions,Mathematics | Journal |
Volume | ISSN | Citations |
317 | 0925-2312 | 2 |
PageRank | References | Authors |
0.35 | 36 | 3 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Qiaozhi Zhang | 1 | 2 | 0.35 |
| Di Wang | 2 | 20 | 4.74 |
| Yanguo Wang | 3 | 2 | 0.35 |