A Learned Representation For Artistic Style

Authors: Vincent Dumoulin, Jonathon Shlens, Manjunath Kudlur

ICLR 2017 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental In this work we investigate the construction of a single, scalable deep network that can parsimoniously capture the artistic style of a diversity of paintings. We demonstrate that such a network generalizes across a diversity of artistic styles by reducing a painting to a point in an embedding space. Importantly, this model permits a user to explore new painting styles by arbitrarily combining the styles learned from individual paintings.
Researcher Affiliation Industry Vincent Dumoulin & Jonathon Shlens & Manjunath Kudlur Google Brain, Mountain View, CA vi.dumoulin@gmail.com, shlens@google.com, keveman@google.com
Pseudocode No The paper does not contain a structured pseudocode or algorithm block. Figure 3 shows mathematical equations, not pseudocode.
Open Source Code Yes A complete implementation of the model in Tensor Flow (Abadi et al., 2016) as well as a pretrained model are available for download 1. 1https://github.com/tensorflow/magenta
Open Datasets Yes Our training procedure follows Johnson et al. (2016). Briefly, we employ the Image Net dataset (Deng et al., 2009) as a corpus of training content images.
Dataset Splits No The paper mentions 'evaluation images' but does not specify explicit training, validation, and test dataset splits with percentages or counts.
Hardware Specification No The paper does not provide specific hardware details (e.g., GPU/CPU models, memory) used for running the experiments.
Software Dependencies No The paper mentions 'Tensor Flow' but does not provide a specific version number for the software used in their implementation (e.g., TensorFlow 1.x or 2.x).
Experiment Setup Yes Unless noted otherwise, all style transfer networks were trained using the hyperparameters outlined in the Appendix s Table 1. ... Optimizer Adam (Kingma & Ba, 2014) (α = 0.001, β1 = 0.9, β2 = 0.999), Parameter updates 40,000, Batch size 16, Weight initialization Isotropic gaussian (µ = 0, σ = 0.01)