Dynamic Hypergraph Structure Learning

Authors: Zizhao Zhang, Haojie Lin, Yue Gao

IJCAI 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on four public datasets show better performance compared with the state-of-the-art methods.
Researcher Affiliation Academia Zizhao Zhang, Haojie Lin, Yue Gao BNRist, KLISS, School of Software, Tsinghua University, China. zz-zh14@mails.tsinghua.edu.cn, haojie.lin@outlook.com, gaoyue@tsinghua.edu.cn
Pseudocode Yes Algorithm 1 Dynamic Hypergraph Structure Learning Input: training data set Q, testing data set D, maximal iteration k, and parameters β and λ Output: label projection matrix F and hypergraph structure H 1: Construct the initial hypergraph G = (V, E, W). 2: Initialize the learning rate α. 3: for i = 0 k 1 do 4: Fix H, and update F by Eq. (6). 5: Fix F, and update H by Eq. (8). 6: end for 7: return F, H
Open Source Code No The paper does not provide an explicit statement or link for open-sourcing the code for the described methodology.
Open Datasets Yes To evaluate the performance of the proposed method on 3D shape recognition, we have conducted experiments on the National Taiwan University 3D shape dataset (NTU) [Chen et al., 2003] and the Engineering Shape Benchmark (ESB) [Jayanti et al., 2006]. To validate the proposed DHSL method on gesture recognition, we have conducted experiments on the MSR gesture 3D dataset (MSRGesture3D) [Wang et al., 2012] and a hand gesture dataset collected by Huazhong University of Science and Technology (Gesture3DMotion).
Dataset Splits No The paper describes training and testing data selection processes, but does not explicitly mention a separate validation set or split, nor does it specify how models were selected if no validation set was used.
Hardware Specification No The paper does not provide specific details about the hardware used for running experiments, such as GPU or CPU models.
Software Dependencies No The paper does not provide specific software dependencies with version numbers required to replicate the experiments.
Experiment Setup Yes In these experiments, we empirically set the parameter β and λ as 10 and 1 on both datasets, respectively.