The Coherent Loss Function for Classification

Authors: Wenzhuo Yang, Melvyn Sim, Huan Xu

ICML 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Section 5 reports the experimental results which show that our classification method outperforms the standard SVM when additional constraints are imposed on the decision function.
Researcher Affiliation Academia Wenzhuo Yang A0096049@NUS.EDU.SG Department of Mechanical Engineering, National University of Singapore, Singapore 117576 Melvyn Sim DSCSIMM@NUS.EDU.SG Department of Decision Sciences, National University of Singapore, Singapore 117576 Huan Xu MPEXUH@NUS.EDU.SG Department of Mechanical Engineering, National University of Singapore, Singapore 117576
Pseudocode No The paper describes mathematical formulations and theorems but does not include any structured pseudocode or algorithm blocks.
Open Source Code No The paper does not provide any statement or link indicating the release of open-source code for the methodology described.
Open Datasets Yes Three binary-class datasets Breast cancer , Ionosphere and Diabetes , and two multi-class datasets Wine and Iris from UCI (Asuncion & Newman, 2007) are used, where we randomly pick 50% as training samples, 20% as validation samples, and the rest as testing samples.
Dataset Splits Yes Three binary-class datasets Breast cancer , Ionosphere and Diabetes , and two multi-class datasets Wine and Iris from UCI (Asuncion & Newman, 2007) are used, where we randomly pick 50% as training samples, 20% as validation samples, and the rest as testing samples.
Hardware Specification No The paper states: 'To solve the resulting optimization problems, we use CVX (Grant & Boyd, 2011; 2008), and Gurobi (Gurobi Optimization, 2013) as the solver.' but provides no specific hardware details used for the experiments.
Software Dependencies Yes To solve the resulting optimization problems, we use CVX (Grant & Boyd, 2011; 2008), and Gurobi (Gurobi Optimization, 2013) as the solver.
Experiment Setup Yes For the cumulative loss formulation approach, parameter C is determined by cross-validation. For the coherent loss formulation approach, parameter a is determined by cross-validation. For each T, we repeated the experiments 20 times and computed the average classification errors. To solve the resulting optimization problems, we use CVX (Grant & Boyd, 2011; 2008), and Gurobi (Gurobi Optimization, 2013) as the solver.