Michael Maire

Hierarchical Implicit Neural Emulators

Approaching Deep Learning through the Spectral Dynamics of Weights

Residual Connections Harm Generative Representation Learning

Nested Diffusion Models using Hierarchical Latent Priors

Latent Intrinsics Emerge from Training to Relight

HyperFields: Towards Zero-Shot Generation of NeRFs from Text

Factorized Diffusion Architectures for Unsupervised Image Generation and Segmentation

Structural Adversarial Objectives For Self-Supervised Representation Learning