Improving Semi-Supervised Target Alignment via Label-Aware Base Kernels

Authors: Qiaojun Wang, Kai Zhang, Guofei Jiang, Ivan Maric

AAAI 2014 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental This section compares our method with a number of state-of-the-art algorithms for semi-supervised kernel design, for both classification and regression. ... Results are reported in Table 1.
Researcher Affiliation Collaboration 1Department of Electrical and Computer Engineering, Rutgers University, Piscataway, NJ 08854 USA qjwang@eden.rutgers.edu, marsic@ece.rutgers.edu 2NEC Laboratories America, Inc. 4 Independence Way, Princeton, NJ 08540 USA {kzhang,gfj}@nec-labs.com
Pseudocode Yes Algorithm 1 Input: labeled samples Xl = {xi}l i=1, unlabeled sample set Xu = {xi}n i=l+1; Gaussian Kernel k( , ), label Y = [y1, y2, ..., yc] Rl c.
Open Source Code No No statement explicitly providing open-source code for the methodology, or a link to a code repository, was found.
Open Datasets Yes This section compares our method with a number of state-of-the-art algorithms for semi-supervised kernel design, for both classification and regression. ... Digit1, USPS, COIL2, BSI, COIL, g241n, Text, usps38, usps49, usps56, usps27, odd/even ... The task is indoor location estimation using received signal strength(RSS) that a client device received from Wi-Fi access points (Yang, Pan, and Zheng 2000).
Dataset Splits No No specific training/validation/test dataset splits with exact percentages, sample counts, or citations to predefined validation splits were found. The paper mentions '50 labeled samples randomly chosen for each class' and training/test set usage, but not a distinct validation split.
Hardware Specification No No specific hardware details (e.g., CPU, GPU models, memory) used for running the experiments were provided.
Software Dependencies No The paper mentions 'libsvm package' but does not provide a specific version number. No other software dependencies with version numbers are listed.
Experiment Setup Yes For the kernel width, we first compute b0 as the inverse of the average squared pairwise distances, and then choose b among b0 { 1/5, 1, 5, 10} that gives the best performance. The parameter δ and ϵ are chosen from {10^-5, 10^-3, 10^-1, 1}. ... The regularization parameter C is chosen as{0.1, 1, 10, 100, 1000, 10000}.