Veto-Consensus Multiple Kernel Learning

Authors: Yuxun Zhou, Ninghang Hu, Costas Spanos

AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental The proposed method is evaluated on extensive set of experiments, and the results show significant improvement over the state-of-the-art approaches.
Researcher Affiliation Academia Yuxun Zhou Department of EECS UC Berkeley yxzhou@berkeley.edu Ninghang Hu Amsterdam Machine Learning Lab University of Amsterdam huninghang@gmail.com Costas J. Spanos Department of EECS UC Berkeley spanos@berkeley.edu
Pseudocode Yes Algorithm 1 PDDP for VCMKL
Open Source Code No The paper does not provide an explicit statement about releasing open-source code for the described methodology or a link to a code repository.
Open Datasets Yes The public UCI (Lichman 2013) pima data set is used in this experiment, and the learning problem OPT1 is solved with different algorithms. [...] two data sets (i.e. the UCI robot execution and vowel) are used to demonstrate the effect of number of kernels M on classification performance.
Dataset Splits Yes The other hyperparameters for all methods are chosen with 10 folds CV.
Hardware Specification No The paper does not provide specific details about the hardware used to run the experiments, such as CPU/GPU models or memory specifications.
Software Dependencies No The paper does not provide specific software dependencies with version numbers used for replicating the experiments.
Experiment Setup Yes Hyperparameters are set with M = 5, νm = 0.02 m and γ = |I |/l 2. For PDDP initialization a simple K-mean is applied to the negative class and λ1 is assigned according to cluster labels. The final value of the objective function, the corresponding testing accuracy, number of iterations and the time consumed are shown in Table 1, for which |Jt+1 Jt|/Jt 10 4 is chosen as the stopping criteria. [...] an incremental CV is applied for selecting M. The other hyperparameters for all methods are chosen with 10 folds CV.