Search Results for author: Binxu Wang

Found 7 papers, 4 papers with code

The Hidden Linear Structure in Score-Based Models and its Application

no code implementations17 Nov 2023 Binxu Wang, John J. Vastola

We claimed that for well-trained diffusion models, the learned score at a high noise scale is well approximated by the linear score of Gaussian.

On the Level Sets and Invariance of Neural Tuning Landscapes

no code implementations26 Dec 2022 Binxu Wang, Carlos R. Ponce

This shows the power of level sets as a conceptual tool to understand neuronal activations over image space.

High-performance Evolutionary Algorithms for Online Neuron Control

1 code implementation14 Apr 2022 Binxu Wang, Carlos R. Ponce

To find these patterns, we have used black-box optimizers to search a 4096d image space, leading to the evolution of images that maximize neuronal responses.

Evolutionary Algorithms Vocal Bursts Intensity Prediction

On the use of Cortical Magnification and Saccades as Biological Proxies for Data Augmentation

1 code implementation NeurIPS Workshop SVRHM 2021 Binxu Wang, David Mayo, Arturo Deza, Andrei Barbu, Colin Conwell

Critically, we find that random cropping can be substituted by cortical magnification, and saccade-like sampling of the image could also assist the representation learning.

Data Augmentation Representation Learning +1

The Geometry of Deep Generative Image Models and its Applications

1 code implementation15 Jan 2021 Binxu Wang, Carlos R. Ponce

Our results illustrate that defining the geometry of the GAN image manifold can serve as a general framework for understanding GANs.

Image-Variation

A Geometric Analysis of Deep Generative Image Models and Its Applications

1 code implementation ICLR 2021 Binxu Wang, Carlos R Ponce

We show that the use of this metric allows for more efficient optimization in the latent space (e. g. GAN inversion) and facilitates unsupervised discovery of interpretable axes.

Image-Variation

Cannot find the paper you are looking for? You can Submit a new open access paper.