Do we need hundreds of classifiers to solve real world classification problems?

Paper

Datasets

Detailed results

Complete results (Accuracy)

Complete results (Kappa)

Direct Kernel Perceptron (DKP): Please reference the paper (Bibtex):

@article{delgado14,
    author = {M. Fern\'andez-Delgado, E. Cernadas, S. Barro, J. Ribeiro, J. Neves},
    title = {Direct Kernel Perceptron (DKP): ultra-fast kernel ELM-based classification with non-iterative closed-form weight calculation},
    journal = {Neural Networks},
}

Programs:

• C code

• Matlab code

Direct Parallel Perceptrons (DPPs): Please reference the paper (Bibtex):

@article{delgado11,
    author = {M. Fern\'andez-Delgado, J. Ribeiro, E. Cernadas, S. Barro},
    title = {Direct Parallel Perceptrons (DPPs): fast analytical calculation of the Parallel Perceptron weights with margin control for classification tasks},
    journal = {IEEE Trans. On Neural Networks},
    volume = {22},
    pages = {1837—1848},
    year = {2011}

}

Programs:

1. dpp_two_class.c: 2-Class Direct Parallel Perceptron (DPP)

2. dpp_multi_class.c: Multi-Class Direct Parallel Perceptron (DPP)

3. dpp_matlab_implementation.zip: