Papers2

#CIFAR-10

Image Generation with a Sphere Encoder

Kaiyu Yue, Menglin Jia et al.Feb 16arXiv

The Sphere Encoder is a new way to make images fast by teaching an autoencoder to place all images evenly on a big imaginary sphere and then decode random spots on that sphere back into pictures.

#Sphere Encoder#Spherical Latent Space#RMS Normalization

Not triaged yet

Visualizing the Loss Landscape of Neural Nets

Intermediate

Hao Li, Zheng Xu et al.Dec 28arXiv

Training a neural network is like finding the lowest spot in a giant, bumpy landscape called the loss landscape.

#loss landscape visualization#filter normalization#sharpness flatness

Not triaged yet