McGan: Mean and Covariance Feature Matching GAN

Authors: Youssef Mroueh, Tom Sercu, Vaibhava Goel

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

Reproducibility Variable Result LLM Response
Research Type Experimental We train Mc Gan for image generation with both Mean Matching and Covariance Matching objectives. We show generated images on the labeled faces in the wild (lfw), LSUN bedrooms, and cifar-10 datasets.
Researcher Affiliation Industry Youssef Mroueh * 1 2 Tom Sercu * 1 2 Vaibhava Goel 2 1AI Foundations. IBM T.J. Watson Research Center, NY, USA 2Watson Multimodal Algorithms and Engines Group. IBM T.J. Watson Research Center, NY, USA. Correspondence to: Youssef Mroueh <mroueh@us.ibm.com>.
Pseudocode Yes Algorithm 1 Mean Matching GAN Primal (Pµ) Algorithm 2 Mean Matching GAN Dual (Dµ) Algorithm 3 Covariance Matching GAN Primal (PΣ)
Open Source Code No The paper mentions using 'the code released with WGAN (Arjovsky et al., 2017)' but does not state that the authors' own code for the methodology described is publicly available or open source.
Open Datasets Yes We train Mc Gan for image generation with both Mean Matching and Covariance Matching objectives. We show generated images on the labeled faces in the wild (lfw) (Huang et al., 2007), LSUN bedrooms (Yu et al., 2015), and cifar-10 (Krizhevsky & Hinton, 2009) datasets.
Dataset Splits No The paper does not explicitly provide training/validation/test dataset splits by percentage or sample counts, nor does it specify a cross-validation setup for reproducibility.
Hardware Specification No The paper does not explicitly describe the hardware used to run its experiments, such as specific GPU/CPU models, memory, or cloud computing instances.
Software Dependencies No The paper mentions following 'DCGAN principles', using 'batch normalization' and 'Re LU activations', and 'RMSProp' optimizer, but does not provide specific version numbers for any software components, libraries, or frameworks.
Experiment Setup Yes We generate 64 64 images for lfw and LSUN and 32 32 images on cifar, and train with minibatches of size 64. We follow the experimental framework and implementation of (Arjovsky et al., 2017), where we ensure the boundedness of Φω by clipping the weights pointwise to the range [ 0.01, 0.01]. We observed that the default learning rates from WGAN (5e-5) are optimal for both primal and dual formulation. For the dual formulation (Algorithm 2), we confirmed the hypothesis that we need a good estimate of µω(Pr) in order to compute the gradient of the generator θ: we needed to increase the minibatch size of real threefold to 3 64. We use Algorithm 3 with k = 16 components. The critic loss becomes LD = ˆ Lσ λD 1 (xi,yi) lab CE(xi, yi; S, Φω), with hyper-parameter λD. The generator loss (with hyper-param λG) becomes: LG = ˆ Lσ +λG 1 zi pz,yi py CE(gθ(zi, yi), yi; S, Φω).