Pandaemonium Architecture 6.0

In this course

Salimans et al. are cited here as early diagnosticians of failure modes in GAN training, specifically the problem of texture blobs rather than coherent objects appearing in generated images. Their work establishes a baseline understanding of what goes wrong in generative models, against which the paper's own contributions (like those in Brock et al.'s high-fidelity synthesis) can be measured and distinguished. This positions them as part of the foundational critical vocabulary for evaluating generative adversarial networks in the history of deep learning.

Mentioned in 1 reading

Andrew Brock

et al., LARGE SCALE GAN TRAINING FOR HIGH FIDELITY NATURAL IMAGE SYNTHESIS

4

Large Scale GAN Training for High Fidelity Natural Image Synthesis Andrew Brock\† Heriot-Watt University ajb5@hw.ac.uk Jeff Donahue† DeepMind jeffdonahue@google.com Karen Simonyan† DeepMind simonyan@google.com Abstract Despite recent p…A Generative Adversarial Network (GAN) involves Generator (G) and Discriminator (D) networks whose purpose, respectively, is to map random noise to samples and discriminate real and generated samples. Formally, the GAN objective, in its ori…Dumoulin et al. (2017) modify the way class conditioning is passed to G by supplying it with class-conditional gains and biases in BatchNorm (Ioffe & Szegedy, 2015) layers. In Miyato & Koyama (2018), D is conditioned by using the cosine sim…

+1 more

Appears alongside

People mentioned in the same passages — sorted by co-occurrence weight.