Top-k Multiclass SVM

Authors: Maksim Lapin, Matthias Hein, Bernt Schiele

NeurIPS 2015 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on five datasets show consistent improvements in top-k accuracy compared to various baselines. Finally, extensive experiments on several challenging computer vision problems show that top-k multiclass SVM consistently improves in top-k error over the multiclass SVM (equivalent to our top-1 multiclass SVM), one-vs-all SVM and other methods based on different ranking losses [11, 16].
Researcher Affiliation Academia 1Max Planck Institute for Informatics, Saarbrücken, Germany 2Saarland University, Saarbrücken, Germany
Pseudocode Yes Algorithm 1 Top-k Multiclass SVM
Open Source Code Yes We release our implementation of the projection procedures and both SDCA solvers as a C++ library2 with a Matlab interface. 2https://github.com/mlapin/libsdca
Open Datasets Yes We evaluate our method on five image classification datasets of different scale and complexity: Caltech 101 Silhouettes [26] (m = 101, n = 4100), MIT Indoor 67 [20] (m = 67, n = 5354), SUN 397 [29] (m = 397, n = 19850), Places 205 [30] (m = 205, n = 2448873), and Image Net 2012 [22] (m = 1000, n = 1281167).
Dataset Splits Yes We cross-validate hyper-parameters in the range 10−5 to 103, extending it when the optimal value is at the boundary.
Hardware Specification No No specific hardware details (like CPU/GPU models, memory, or cluster specifications) used for running experiments are explicitly stated.
Software Dependencies No The paper mentions several software tools and libraries (e.g., Lib Linear, SVMPerf, Caffe), but no specific version numbers are provided for any of them.
Experiment Setup No The paper states, 'We cross-validate hyper-parameters in the range 10−5 to 103', but does not provide specific hyperparameter values (e.g., learning rate, batch size, epochs) used for the final models or other detailed training configurations. It references external tools, which would imply their default or documented settings.