Clustering Approach to Solve Hierarchical Classification Problem Complexity

Authors: Aomar Osmani, Massinissa Hamidi, Pegah Alizadeh7904-7912

AAAI 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Extensive experiments on the SHL dataset show that our approach improves classification performances while reducing the number of instances used to learn each concept.
Researcher Affiliation Academia 1 LIPN-UMR CNRS 7030, Univ. Sorbonne Paris Nord 2 De Vinci Research Center, Pˆole Universitaire L eonard de Vinci
Pseudocode Yes Algorithm 1: compute Hierarchy and Algorithm 2: Hierarchy training
Open Source Code Yes Software package and code to reproduce empirical results is publicly available and can be found at https://github.com/sensorrich/clustering-based-HL
Open Datasets Yes We use in our experiments, primarily, the SHL dataset which consists of motion sensor data. ... The University of Sussex-Huawei locomotion and transportation dataset for multimodal analytics with mobile devices. (Gjoreski et al. 2018)
Dataset Splits Yes We use the meta-segmented cross-validation (Hammerla and Pl otz 2015) for model evaluation to alleviate the problem of neighborhood bias and performance over-estimation.
Hardware Specification No The paper does not provide specific hardware details such as GPU models, CPU types, or memory used for running the experiments. It only mentions that 'All training details, hyperparameters, and their sensitivity analysis can be found in the code repository and supplementary materials'.
Software Dependencies No The paper mentions that a 'software package and code to reproduce empirical results is publicly available' and that 'All training details, hyperparameters, and their sensitivity analysis can be found in the code repository and supplementary materials', but it does not specify any software dependencies with version numbers in the main text.
Experiment Setup No The paper states, 'All training details, hyperparameters, and their sensitivity analysis can be found in the code repository and supplementary materials (see C and D)', but does not provide specific hyperparameter values or detailed system-level training settings in the main text.