Learning With Single-Teacher Multi-Student

Authors: Shan You, Chang Xu, Chao Xu, Dacheng Tao

AAAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental Results In this part, we experimentally investigate the effectiveness and efficiency of the proposed gated SVM in dealing with the STMS problem.
Researcher Affiliation Academia Key Lab. of Machine Perception (MOE), Cooperative Medianet Innovation Center, School of EECS, Peking University, China UBTECH Sydney AI Centre, SIT, FEIT, University of Sydney, Australia youshan@pku.edu.cn, c.xu@sydney.edu.au xuchao@cis.pku.edu.cn, dacheng.tao@sydney.edu.au
Pseudocode Yes Algorithm 1 Difficulty Coding with Single-Teacher and Algorithm 2 Training for gated SVM (g SVM) with Majorization Minimization optimization
Open Source Code No The paper does not include any explicit statement or link indicating the release of its source code for the methodology described.
Open Datasets Yes Datasets. We used 6 benchmark multi-class datasets in our experiment, including letter (Hsu and Lin 2002), protein (Wang 2002), satimage (Hsu and Lin 2002), shuttle (Hsu and Lin 2002), vowel and yeast.2 The various statistics of all datasets are presented in Tabel 2. 2Datasets can be downloaded in https://www.csie.ntu.edu.tw/ cjlin/libsvmtools/datasets/multiclass.html
Dataset Splits Yes The parameters C, ε and ϵ are tuned by 10-fold cross validation in the sets {10 3, ..., 100, ..., 103}, {0, 0.1, 0.2, ..., 1.5} and {0.1, 0.2, ..., 0.5}, respectively.
Hardware Specification No The paper does not provide specific hardware details (such as CPU/GPU models, memory, or cloud instance types) used for running the experiments.
Software Dependencies No The paper mentions 'mosek (Mosek 2010) and cvx (Grant, Boyd, and Ye 2008)' as toolboxes, but does not provide specific version numbers for any software dependencies used in their experimental setup.
Experiment Setup Yes The parameters C, ε and ϵ are tuned by 10-fold cross validation in the sets {10 3, ..., 100, ..., 103}, {0, 0.1, 0.2, ..., 1.5} and {0.1, 0.2, ..., 0.5}, respectively.