Multiclass Support Vector Data Description in Extreme Learning Machine Feature Space

Abstract Support vector data description (SVDD) method aims to address the one-class classification (OCC) problem find a hypersphere-shaped of target set. For extending SVDD multiclass while remaining ability detecting outliers, we propose novel scheme which can be used in effective feature mapping and meta-class separation based on extreme learning machine algorithm (ELM-MSVDD). Accordingly, imprecise difficult distinguish specific classes is classified meta-class,the defined by disjunction these classes, this operation reduce error rate effectively. Experimental results our ELM-MSVDD show well performance datasets from UCI library radar signal source recognition. Meanwhile, proposed provide theoretical practical support for other relevant pattern recognition field.