Multi-Layer Multi-View Classification for Alzheimer’s Disease Diagnosis

Authors: Changqing Zhang, Ehsan Adeli, Tao Zhou, Xiaobo Chen, Dinggang Shen

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results on Alzheimers Disease Neuroimaging Initiative (ADNI) data sets validate the effectiveness of our proposed method.
Researcher Affiliation Academia a Department of Radiology and BRIC, University of North Carolina at Chapel Hill, North Carolina, USA b School of Computer Science and Technology, Tianjin University, Tianjin, China c Department of Psychiatry and Behavioral Sciences & Stanford AI Lab (SAIL), Stanford University, California, USA
Pseudocode Yes Algorithm 1: Optimization for our ML-MVC model.
Open Source Code No No statement about making the source code publicly available for the described methodology was found.
Open Datasets Yes Experimental results on Alzheimers Disease Neuroimaging Initiative (ADNI) data sets validate the effectiveness of our proposed method. ... we download ADNI 1.5T MR and PET images from the ADNI website 2. ... 2http://adni.loni.usc.edu/
Dataset Splits Yes In all experiments, the data are split into 10 non-overlapping folds with 9/10 and 1/10 as training and testing data, and reporting the average results and standard deviation. We conduct standard 10-fold crossvalidation for each split with the hyperparameters selected from {0.01, 0.1, 1, 10, 100} for α, and {0.1, 1, 10, 100} for the other hyperparameters.
Hardware Specification No No specific details about the hardware (e.g., CPU/GPU models, memory, or specific computing environments) used for running the experiments were provided.
Software Dependencies No We employ a support vector classification model as the basic classifier which is from the LIBSVM toolbox 1 publicly available for the compared methods.
Experiment Setup Yes We conduct standard 10-fold crossvalidation for each split with the hyperparameters selected from {0.01, 0.1, 1, 10, 100} for α, and {0.1, 1, 10, 100} for the other hyperparameters. Gaussian kernel is employed for each type of features, i.e., k(xi, xj) = φ(xi), φ(xj) = exp( ||xi xj||2 / 2σ2 ) where σ = median({||xi xj||}i =j). Initialize: P(1) = = P(V ) = 0, G = K = W = 0, ρ = 1.2, ϵ = 10 6, maxμ =106; Initialize S with random values.