Local Discriminant Hyperalignment for Multi-Subject fMRI Data Alignment
Authors: Muhammad Yousefnezhad, Daoqiang Zhang
AAAI 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental studies on multi-subject MVP analysis confirm that the LDHA method achieves superior performance to other state-of-the-art HA algorithms. |
| Researcher Affiliation | Academia | Muhammad Yousefnezhad, Daoqiang Zhang College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing, China. {myousefnezhad, dqzhang}@nuaa.edu.cn |
| Pseudocode | Yes | Algorithm 1 Local Discriminate Hyperalignment (LDHA) and Algorithm 2 A general template for MVP analysis by using Local Discriminate Hyperalignment (LDHA). |
| Open Source Code | No | The paper states, 'All algorithms are implemented in the MATLAB R2016b (9.1) on a PC with certain specifications1 by authors in order to generate experimental results.' but does not provide a link or explicit statement about making their code open source. |
| Open Datasets | Yes | This paper utilizes 4 datasets, shared by openfmri.org, for running empirical studies of this section... As the first dataset, Visual Object Recognition (DS105) includes... Please see (Haxby et al. 2011; Haxby, Connolly, and Guntupalli 2014) for more information... As the second dataset, Multisubject, multi-modal human neuroimaging dataset (DS117)... Please see (Wakeman and Henson 2015) for more information... As the third dataset, Word and Object Processing (DS107)... ROIs and technical information are defined based on (Duncan et al. 2009)... As the last dataset, Mixed-gambles task (DS005)... Further, the ROIs for functional alignment are selected based on the original paper (Tom et al. 2007)... As the first dataset, A high-resolution 7-Tesla f MRI dataset from complex natural stimulation with an audio movie (DS113) includes the f MRI data of 20 subjects, who watched Forrest Gump (1994) movie during the experiment. This dataset provided by www.openfmri.org. Please see (Hanke et al. 2014) for more information. |
| Dataset Splits | Yes | The features (voxels in the ROIs) are partitioned to train set and test set by using Leave-One-Out (LOO) crossvalidation across subjects (leave-one-subject-out). |
| Hardware Specification | Yes | 1DEL , CPU = Intel Xeon E5-2630 v3 (8 2.4 GHz), RAM = 64GB, OS = Elementary OS 0.4 Loki |
| Software Dependencies | Yes | All algorithms are implemented in the MATLAB R2016b (9.1) |
| Experiment Setup | No | The paper states that classification models were generated using ν-SVM algorithms (binary and multi-label) and that KHA employed a Gaussian kernel, with SCCA parameters considered 'optimum based on (Xu et al. 2012)'. However, it does not provide specific hyperparameter values (e.g., for ν-SVM, or the specific 'optimum' values for SCCA) needed to reproduce the experimental setup. |