Back to Search
Start Over
Least absolute deviation-based robust support vector regression
- Source :
- Knowledge-Based Systems. 131:183-194
- Publication Year :
- 2017
- Publisher :
- Elsevier BV, 2017.
-
Abstract
- To suppress the influence of outliers on function estimation, we propose a least absolute deviation (LAD)-based robust support vector regression (SVR). Furthermore, an efficient algorithm based on the split-Bregman iteration is introduced to solve the optimization problem of the proposed algorithm. Both artificial and benchmark datasets are employed to compare the performance of the proposed algorithm with those of least squares SVR (LS-SVR), and two weighted versions of LS-SVR with the weight functions of Hampel and Logistic, respectively. Experiments demonstrate the superiority of the proposed algorithm.
- Subjects :
- 0209 industrial biotechnology
Mathematical optimization
Information Systems and Management
Optimization problem
Computer science
02 engineering and technology
Function (mathematics)
Least squares
Management Information Systems
Robust regression
Support vector machine
020901 industrial engineering & automation
Artificial Intelligence
Outlier
0202 electrical engineering, electronic engineering, information engineering
Benchmark (computing)
020201 artificial intelligence & image processing
Least absolute deviations
Algorithm
Software
Subjects
Details
- ISSN :
- 09507051
- Volume :
- 131
- Database :
- OpenAIRE
- Journal :
- Knowledge-Based Systems
- Accession number :
- edsair.doi...........a0194e44c14ea17acd59858e328147bc