Exploiting Task-Feature Co-Clusters in Multi-Task Learning

Authors: Linli Xu, Aiqing Huang, Jianhui Chen, Enhong Chen

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

Reproducibility Variable Result LLM Response
Research Type Experimental In this section, we evaluate the proposed approach of multitask learning with task-feature co-clusters (Co CMTL) in comparison with single task learning methods as well as representative multi-task learning algorithms. The experiments are first conducted in a synthetic setting, and then on two real-world data sets.
Researcher Affiliation Collaboration School of Computer Science and Technology University of Science and Technology of China, Hefei, Anhui 230027, China Yahoo Labs, Sunnyvale, CA 94089, USA
Pseudocode Yes Algorithm 1 Iterative algorithm for computing Qr
Open Source Code No The paper mentions "All algorithms are implemented with MATLAB." but does not provide any concrete access to the source code for the methodology described, nor does it state that the code is open-source or available.
Open Datasets Yes The School data consists of the exam scores of 15362 students from 139 secondary schools in London during the years of 1985-1987; each student is described by 27 attributes including gender, ethnic group, etc. Here we follow the experimental setup as in (Chen, Zhou, and Ye 2011)
Dataset Splits Yes 10%, 20%, and 30% of the samples from each task are randomly selected as training sets and the rest are used as test sets.
Hardware Specification No The paper does not provide specific hardware details such as exact GPU/CPU models, processor types, or memory amounts used for running its experiments. It only mentions that algorithms were implemented in MATLAB.
Software Dependencies No The paper states "All algorithms are implemented with MATLAB." but does not provide specific version numbers for MATLAB or any other software dependencies needed to replicate the experiment.
Experiment Setup Yes In the experiments, the hyper-parameters are tuned by 3-fold cross validation. All algorithms are implemented with MATLAB. The maximum number of iterations is set to 5000 for all algorithms, with tolerance of 10 5.