Binary Classification from Multiple Unlabeled Datasets via Surrogate Set Classification

Authors: Nan Lu, Shida Lei, Gang Niu, Issei Sato, Masashi Sugiyama

ICML 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Through experiments, we demonstrate the superiority of our proposed method over state-of-the-art methods.
Researcher Affiliation Academia 1The University of Tokyo, Tokyo, Japan 2RIKEN, Tokyo, Japan.
Pseudocode Yes Algorithm 1 Um-SSC based on stochastic optimization
Open Source Code Yes Our implementation of Um-SSC is available at https://github.com/leishida/Um-Classification.
Open Datasets Yes Datasets We train on widely adopted benchmarks MNIST, Fashion-MNIST, Kuzushiji-MNIST, and CIFAR-10.
Dataset Splits No The paper mentions 'training data' and 'test phase' but does not explicitly provide details for a distinct validation set or its split.
Hardware Specification No The paper does not specify the hardware used for running the experiments (e.g., GPU models, CPU types).
Software Dependencies No The paper mentions using 'Adam (Kingma & Ba, 2015) with the cross-entropy loss for optimization' but does not specify version numbers for Adam, the specific deep learning framework (e.g., PyTorch, TensorFlow), or other software dependencies.
Experiment Setup Yes We train 300 epochs for all the experiments, and the classification error rates at the test phase are reported. All the experiments are repeated 3 times and the mean values with standard deviations are recorded for each method.