Self-supervised Label Augmentation via Input Transformations

Authors: Hankook Lee, Sung Ju Hwang, Jinwoo Shin

ICML 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We experimentally validate our self-supervised label augmentation techniques described in Section 2. Throughout this section, we refer to data augmentation LDA (2) as DA, multi-task learning LMT (1) as MT, and our self-supervised label augmentation LSLA (3) as SLA for notational simplicity.
Researcher Affiliation Collaboration 1School of Electrical Engineering, KAIST, Daejeon, Korea 2Graduate School of AI, KAIST, Daejeon, Korea 3School of Computing, KAIST, Daejeon, Korea 4AITRICS, Seoul, Korea.
Pseudocode Yes In the supplementary material, we provide pseudo-codes of our algorithm, which can be easily implemented.
Open Source Code Yes Code available at https://github.com/hankook/SLA.
Open Datasets Yes We evaluate our method on various classification datasets: CIFAR10/100 (Krizhevsky et al., 2009), Caltech-UCSD Birds or CUB200 (Wah et al., 2011), Indoor Scene Recognition or MIT67 (Quattoni & Torralba, 2009), Stanford Dogs (Khosla et al., 2011), and tiny-Image Net3 for standard or imbalanced image classification; mini-Image Net (Vinyals et al., 2016), CIFAR-FS (Bertinetto et al., 2019), and FC100 (Oreshkin et al., 2018) for few-shot classification.
Dataset Splits No The paper describes training datasets and test sets, but does not explicitly provide details about validation dataset splits (e.g., percentages, sample counts, or specific files/citations for validation sets).
Hardware Specification No The paper does not provide specific details about the hardware (e.g., GPU models, CPU types, or memory) used for running the experiments.
Software Dependencies No The paper mentions various methods and models (e.g., SGD, ResNet-32, Meta Opt Net) but does not provide specific software dependencies with version numbers (e.g., programming language versions, library versions, or framework versions).
Experiment Setup Yes For the standard image classification datasets, we use SGD with a learning rate of 0.1, momentum of 0.9, and weight decay of 10 4. We train for 80k iterations with a batch size of 128. For the finegrained datasets, we train for 30k iterations with a batch size of 32 because they have a relatively smaller number of training samples. We decay the learning rate by the constant factor of 0.1 at 50% and 75% iterations.