Unsupervised Representation Learning by Predicting Image Rotations

Authors: Spyros Gidaris, Praveer Singh, Nikos Komodakis

ICLR 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We exhaustively evaluate our method in various unsupervised feature learning benchmarks and we exhibit in all of them state-of-the-art performance.
Researcher Affiliation Academia Spyros Gidaris, Praveer Singh, Nikos Komodakis University Paris-Est, LIGM Ecole des Ponts Paris Tech {spyros.gidaris,praveer.singh,nikos.komodakis}@enpc.fr
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code No The code and models of our paper will be published on: https://github.com/gidariss/Feature Learning Rot Net. This indicates a future release, not immediate concrete access.
Open Datasets Yes In this section we conduct an extensive evaluation of our approach on the most commonly used image datasets, such as CIFAR-10 (Krizhevsky & Hinton, 2009), Image Net (Russakovsky et al., 2015), PASCAL (Everingham et al., 2010), and Places205 (Zhou et al., 2014)
Dataset Splits No The paper mentions 'training' and 'test' sets for datasets like CIFAR-10 and ImageNet, but it does not explicitly describe a 'validation split' or its size/methodology from the overall dataset for hyperparameter tuning or model selection.
Hardware Specification Yes our Alex Net model trains in around 2 days using a single Titan X GPU
Software Dependencies No The paper describes the neural network architectures and training optimizers (e.g., SGD, batch-norm, relu units) but does not list specific software libraries or their version numbers (e.g., PyTorch, TensorFlow, etc.).
Experiment Setup Yes In order to train them on the rotation prediction task, we use SGD with batch size 128, momentum 0.9, weight decay 5e 4 and lr of 0.1. We drop the learning rates by a factor of 5 after epochs 30, 60, and 80. We train in total for 100 epochs.