Teaching an Active Learner with Contrastive Examples

Authors: Chaoqi Wang, Adish Singla, Yuxin Chen

NeurIPS 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental 5 Numerical Experiments. We conducted experiments on two datasets
Researcher Affiliation Academia Chaoqi Wang University of Chicago chaoqi@uchicago.edu Adish Singla MPI-SWS adishs@mpi-sws.org Yuxin Chen University of Chicago chenyuxin@uchicago.edu
Pseudocode Yes Algorithm 2 Greedy teaching algorithm for active leaner
Open Source Code No The paper does not include any explicit statement about providing open-source code for the described methodology, nor does it provide a link to a code repository.
Open Datasets Yes We conducted experiments on two datasets: Vespula-Weevil and Butterfly-Moth, which were adopted by Singla et al. [2013, 2014] for verifying the algorithm for teaching the crowd to classify images.
Dataset Splits No The paper describes the datasets and how they were used, but it does not specify explicit training, validation, or test dataset splits.
Hardware Specification No The paper does not specify any hardware details (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper does not list any specific software dependencies with version numbers.
Experiment Setup Yes 5.1 Experimental Setup: β-greedy active learners: In the experiments, we consider the class of β-greedy active learners, for which the selected query xq t+1 satisfies ... We can interpolate β from 1 to + to cover a broad range of active learners with different properties. ... In terms of the constraint on the teacher’s example, we consider the following three types of the constraints: 1) examples that are close to the learner’s query but with different label (denoted by C); 2) examples that are far away from the learner’s query but with the same label(denoted by F); 3) the union of 1)& 2) (denoted by C+F). For each constraint, we also adopt a parameter ψ [0, 1] to control the size of the search space...