Finding the Global Semantic Representation in GAN through Fréchet Mean

Authors: Jaewoong Choi, Geonho Hwang, Hyunsoo Cho, Myungjoo Kang

ICLR 2023 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experimental results demonstrate that Fr echet basis provides better semantic factorization and robustness compared to the previous methods. Moreover, we suggest the basis refinement scheme for the previous methods. The quantitative experiments show that the refined basis achieves better semantic factorization while constrained on the same semantic subspace given by the previous method.
Researcher Affiliation Academia Jaewoong Choi Korea Institute for Advanced Study chjw1475@kias.re.kr Geonho Hwang, Hyunsoo Cho, Myungjoo Kang Seoul National University {hgh2134,hscho100,mkang}@snu.ac.kr
Pseudocode No The paper does not contain any clearly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not provide an explicit statement or link for open-source code for the described methodology.
Open Datasets Yes We evaluate the Fr echet basis as the global semantic perturbations on the intermediate layers of the mapping network in various Style GAN models (Karras et al., 2019; 2020b)... We utilized 40 binary attribute classifiers pre-trained on Celeb A (Liu et al., 2015) to annotate the 10k generated images... The image traversals are performed on Style GAN2-FFHQ (Fig 3a, 3b) and Style GAN2-LSUN cat (Fig 3c, 3d).
Dataset Splits No The paper mentions training models and evaluating them, but it does not explicitly specify the train/validation/test split percentages or sample counts for the datasets used.
Hardware Specification No The paper does not provide specific details about the hardware used for running the experiments (e.g., GPU models, CPU types).
Software Dependencies No The paper mentions "Pymanopt (Townsend et al., 2016)" but does not provide a specific version number for it or any other software components.
Experiment Setup Yes In this section, all Fr echet basis are discovered using 1,000 i.i.d. samples of Local Basis with θpre = 0.01. The max iteration is set to 200 when optimizing Fr echet mean with Pymanopt. The perturbation intensity is 2 in Style GAN1, and 5 in Style GAN2-e and Style GAN2.