Exploiting Competition Relationship for Robust Visual Recognition

Authors: Liang Du, Haibin Ling

AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In both experiments our approach demonstrates promising performance gains by exploiting the between-task competition.
Researcher Affiliation Academia Center for Data Analytics and Biomedical Informatics Department of Computer and Information Science Temple University Philadelphia, PA, 19122, USA {liang.du, hbling}@temple.edu
Pseudocode Yes Algorithm 1 Comp Boost
Open Source Code Yes We will release the source code for the Comp Boost algorithm at http://www.dabi.temple.edu/ hbling/code/competingtask.htm.
Open Datasets Yes For actor independent expression recognition, we use the Japanese Female Facial Expression (JAFFE) database (Lyons et al. 1998) as our testbed.
Dataset Splits Yes Five-fold cross validation strategy is applied in the training data to automatically determining the parameter λ in (4) in the candidate set of {λ = m 10k : k { 2, 1, 0}, m {1, 2, 5, 8}}.
Hardware Specification No The paper does not provide specific details about the hardware used for running the experiments.
Software Dependencies No The paper mentions 'Open CV implementation of the Viola-Jones face detector' but does not specify any version numbers for software dependencies.
Experiment Setup Yes In all experiments, we use 100 weak learners in our algorithm, i.e., M = 100. Five-fold cross validation strategy is applied in the training data to automatically determining the parameter λ in (4) in the candidate set of {λ = m 10k : k { 2, 1, 0}, m {1, 2, 5, 8}}.