Active Learning from Crowds with Unsure Option

Authors: Jinhong Zhong, Ke Tang, Zhi-Hua Zhou

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental studies on simulated and real crowdsourcing data show that, by exploiting the unsure option, ALCU-SVM achieves very promising performance.
Researcher Affiliation Academia 1UBRI, School of Computer Science and Technology University of Science and Technology of China, Hefei 230027, China 2National Key Laboratory for Novel Software Technology, Nanjing University, Nanjing 210023, China
Pseudocode Yes Algorithm 1: ALCU-SVM
Open Source Code Yes 1http://staff.ustc.edu.cn/ ketang/codes/IJCAI15ALCU.html
Open Datasets Yes Empirical studies on three UCI data sets (including Pima, Heart and Ionosphere) and a real crowdsourcing data set have been conducted. ... 2https://wiki.cites.illinois.edu/wiki/display/forward/Dataset UDI-Twitter Crawl-Aug2012 [Li et al., 2012]
Dataset Splits Yes Each data set was randomly divided into three parts: initial set, active learning set and testing set. To be specific, the three data sets were divided as: Pima (20,548,200); Heart (20,150,100) and Ionosphere (20,180,151), where the three elements in the parenthesis are the numbers of instances in initial set, active learning set, and testing set respectively.
Hardware Specification No The paper does not provide specific details about the hardware used for running experiments, such as CPU or GPU models, or memory specifications.
Software Dependencies No The text body of each tweet was pre-processed and transferred into a TF-IDF vector by the natural language toolkit (NLTK3)...
Experiment Setup Yes If not specified explicitly, the linear kernel was employed in the classifier (K(x, y) = x y) and the RBF kernel was in reliability models (Kt(x, y) = exp( ||x y||2)).