Binary Decomposition: A Problem Transformation Perspective for Open-Set Semi-Supervised Learning

Authors: Jun-Yi Hang, Min-Ling Zhang

ICML 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Comprehensive experiments on diversified benchmarks clearly validate the superiority of BDMatch as well as the effectiveness of our binary decomposition strategy.
Researcher Affiliation Academia 1School of Computer Science and Engineering, Southeast University, Nanjing 210096, China 2Key Laboratory of Computer Network and Information Integration (Southeast University), Ministry of Education, China. Correspondence to: Min-Ling Zhang <zhangml@seu.edu.cn>.
Pseudocode No The paper describes the BDMatch approach using textual explanations and mathematical equations, but it does not include any pseudocode or a clearly labeled algorithm block.
Open Source Code Yes 2Code package of BDMatch is publicly available at: http: //palm.seu.edu.cn/zhangml/files/BDMatch.rar.
Open Datasets Yes CIFAR-10, CIFAR-100 (Krizhevsky et al., 2009), and Image Net (Deng et al., 2009) with different numbers of labeled data and diversified open-set settings.
Dataset Splits No The paper defines a 'labeled set' and an 'unlabeled set' for training, and an 'evaluation set' for testing, but it does not specify a separate 'validation set' or a distinct validation split for hyperparameter tuning or early stopping.
Hardware Specification No The paper does not specify any particular hardware used for experiments, such as GPU or CPU models, only mentioning model architectures like Wide Res Net-28-2 and Res Net-18.
Software Dependencies No The paper mentions using a 'unified codebase based on USB (Wang et al., 2022b)' and 'SGD optimizer' but does not provide specific version numbers for programming languages, libraries, or other software dependencies.
Experiment Setup Yes The scaling parameter τ controlling the strength of balance is set as 0.5 and the momentum factor µ in Eq.(10) is set as 0.999. Models are all trained by the SGD optimizer and the learning rate is set as 0.03 with a cosine decay. For CIFAR experiments, models are trained for 256 * 1024 iterations and each iteration contains a batch of 64 labeled samples and 64 * 7 unlabeled samples. For Image Net experiments, models are trained for 100 * 1024 iterations and each iteration contains a batch of 32 labeled samples and 32 unlabeled samples. The threshold ρ is set to 0.99 in this paper.