Integrating Bayesian and Discriminative Sparse Kernel Machines for Multi-class Active Learning

Authors: Weishi Shi, Qi Yu

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

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments conducted over both synthetic and real data and comparison with competitive AL methods demonstrate the effectiveness of the proposed model.
Researcher Affiliation Academia Weishi Shi Rochester Institute of Technology ws7586@rit.edu Qi Yu Rochester Institute of Technology qi.yu@rit.edu
Pseudocode No The paper describes its methods using mathematical equations and prose, but it does not include any clearly labeled pseudocode or algorithm blocks.
Open Source Code Yes The details are summarized in the supplementary materials and the source code is available at [21]. Weishi Shi and Qi Yu. Source Code and Data. https://drive.google.com/drive/folders/1kk50i_Dvg_R8Pdp_B8lbt32rqn9KGTDFB4n?usp=sharing, 2019. [Online; accessed 12-October-2019].
Open Datasets Yes We choose 6 datasets from different domains, as summarized in Table 1, to evaluate the proposed KMC based multi-class sampling model. Table 1: Description of Datasets Dataset #Inst #Attr #Classes Class Distr. Domain Yeast 1484 8 10 Skewed Biology Reuters 10788 5227 75 Skewed News Penstroke 1144 500 26 Even Image Derm 1 800 1391 50 Even Medical Derm 2 868 1554 30 Even Medical Auto-drive 58509 48 11 Even Automobile. Weishi Shi and Qi Yu. Source Code and Data. https://drive.google.com/drive/folders/1kk50i_Dvg_R8Pdp_B8lbt32rqn9KGTDFB4n?usp=sharing, 2019. [Online; accessed 12-October-2019].
Dataset Splits No The paper mentions using 70% of synthetic data for training and 30% for testing, but it does not specify a separate validation split or explicit training/validation/test splits for the real datasets.
Hardware Specification No The paper does not provide any specific details regarding the hardware specifications (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies Yes Thus, the problem can be solved much more efficiently by quadratic solvers boosted by sparse input such as MOSEK [20]. MOSEK Ap S. MOSEK Optimizer API for Python 8.1.0.80, 2019.
Experiment Setup Yes For KMC, unless otherwise specified, parameter C is set to 10 2 and the convergence threshold is set to 10 3. Two important parameters of KMC, the large margin coefficient C and the initial sample size S, are set to 1 and 40 respectively when compared with other AL models.