A Probabilistic Derivation of LASSO and L12-Norm Feature Selections

Authors: Di Ming, Chris Ding, Feiping Nie4586-4593

AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental experimental results on six benchmark datasets, including images and bio-microarray data, show that our proposed flexible feature selection method has an overwhelmed advantage over state-of-the-art algorithms.
Researcher Affiliation Academia Di Ming,a Chris Ding,a Feiping Nieb a Department of Computer Science and Engineering, University of Texas at Arlington, TX 76019, USA b Centre for OPTical Imagery Analysis and Learning, Northwestern Polytechnical University, Xian 710072, China
Pseudocode Yes Algorithm 1 Efficient algorithm for solving the ℓ1,2-norm based feature selection. 1: Input: Data matrix X Rd n, labels Y Rn k. 2: Output: W Rd k, Dj Rd d, j = 1, , k. 3: Set t = 0. 4: Initialize Wt. 5: repeat 6: for each class j {1, , k} do 7: Compute Dj via Eq.20. 8: Compute wt+1 j via Eq.24. 9: end for 10: Set t = t + 1. 11: until Converges
Open Source Code No No explicit statement or link regarding the open-sourcing of the code for the methodology described in this paper.
Open Datasets Yes MNIST2 (Lecun et al. 1998), Bin Alpha3, AT&T4. Each instance is represented by a vector with all the pixel values... Carcinomas (Su et al. 2001), (Yang et al. 2006), Lung (Bhattacharjee et al. 2001), TOX5 (Kwon et al. 2012). ... 3https://cs.nyu.edu/ roweis/data.html 4http://www.cl.cam.ac.uk/research/dtg/attarchive/facedatabase. html 5http://featureselection.asu.edu/datasets.php
Dataset Splits Yes Classifiers: k-nearest neighbor (KNN), support vector machine (SVM), and linear regression (LR) with five-fold cross validation are used to evaluate the performance of feature selection on classification. The average of classification performance on different five folds are reported as the final accuracy.
Hardware Specification No No specific hardware details (like GPU/CPU models, memory) are provided for the experimental setup.
Software Dependencies No LIBSVM (Chang and Lin 2011) is used as practical implementation of SVM, in which the kernel is set as linear and C = 1.
Experiment Setup Yes The parameter k in KNN is set as 3. LIBSVM (Chang and Lin 2011) is used as practical implementation of SVM, in which the kernel is set as linear and C = 1.