Generative Adversarial Networks
Toronto AI and the UofT Computer Science Student Union (CSSU) bring you a night of cutting edge AI — with concepts and demo.
We'll navigate through successful approaches used in practice for stable training of a Generative Adversarial Network in high dimensional spaces (millions of dimensions) based on recent research. This is advanced material, but we'll do our best to provide an intuitive understanding of the concepts.
— Spectral Normalization
— Hinge Loss
— Pairwise Distance
— Deconvolution vs Resize-Nearest-Neighbor
— Projection Discriminator
— Input Noise
— Cyclic Learning Rates
— Exponential Decay
— ResNet architecture
Presenter: Dave MacDonald from receptiviti.ai