A Deep Learning Approach to Unsupervised Ensemble Learning

Authors: Uri Shaham, Xiuyuan Cheng, Omer Dror, Ariel Jaffe, Boaz Nadler, Joseph Chang, Yuval Kluger

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on various simulated and real-world datasets demonstrate that our proposed DNN approach outperforms other state-of-the-art methods, in particular when the data violates the conditional independence assumption.
Researcher Affiliation Academia Uri Shaham URI.SHAHAM@YALE.EDU Xiuyuan Cheng XIUYUAN.CHENG@YALE.EDU Omer Dror OMERDR@GMAIL.COM Ariel Jaffe ARIEL.JY@GMAIL.COM Boaz Nadler BOAZ.NADLER@WEIZMANN.AC.IL Joseph Chang JOSEPH.CHANG@YALE.EDU Yuval Kluger YUVAL.KLUGER@YALE.EDU
Pseudocode No The paper describes algorithms and procedures in text, but does not include formal pseudocode blocks or algorithms labeled as such.
Open Source Code Yes Our datasets, as well as the scripts used to obtain the reported results are publicly available at https: //github.com/ushaham/RBMpaper 1.
Open Datasets Yes Our datasets, as well as the scripts used to obtain the reported results are publicly available at https: //github.com/ushaham/RBMpaper 1.
Dataset Splits No The paper does not explicitly provide details about train/validation/test splits, only mentions sample size and evaluation on 'test dataset' without specifying percentages or absolute counts for each split.
Hardware Specification No The paper does not provide specific details about the hardware used for running the experiments (e.g., CPU, GPU models, memory, etc.).
Software Dependencies No The paper mentions using 'Weka machine learning software' and 'Geoffrey Hinton’s website' (which implies Matlab scripts), but does not provide specific version numbers for any software dependencies.
Experiment Setup Yes We remark that in our experiments, we observed that RBMs tend to be highly sensitive to hyper-parameter tuning (such as learning rate, momentum, regularization type and penalty), and these hyper-parameters need to be carefully tuned. To obtain a reasonable hyper-parameter setting we found it useful to apply the random configuration sampling procedure, proposed in (Bergstra & Bengio, 2012), and evaluate different models by average loglikelihood approximation, (see, for example, (Salakhutdinov & Murray, 2008) and the corresponding MATLAB scripts in (Salakhutdinov, 2010)).