Jonas Wulff
YOU?
Author Swipe
View article: Do We Still Need Clinical Language Models?
Do We Still Need Clinical Language Models? Open
Although recent advances in scaling large language models (LLMs) have resulted in improvements on many NLP tasks, it remains unclear whether these models trained primarily with general web text are the right tool in highly specialized, saf…
View article: Procedural Image Programs for Representation Learning
Procedural Image Programs for Representation Learning Open
Learning image representations using synthetic data allows training neural networks without some of the concerns associated with real images, such as privacy and bias. Existing work focuses on a handful of curated generative processes whic…
View article: Seeing What a GAN Cannot Generate
Seeing What a GAN Cannot Generate Open
© 2019 IEEE. Despite the success of Generative Adversarial Networks (GANs), mode collapse remains a serious issue during GAN training. To date, little work has focused on understanding and quantifying which modes have been dropped by a mod…
View article: Learning to See by Looking at Noise
Learning to See by Looking at Noise Open
Current vision systems are trained on huge datasets, and these datasets come with costs: curation is expensive, they inherit human biases, and there are concerns over privacy and usage rights. To counter these costs, interest has surged in…
View article: Using latent space regression to analyze and leverage compositionality in GANs
Using latent space regression to analyze and leverage compositionality in GANs Open
In recent years, Generative Adversarial Networks have become ubiquitous in both research and public perception, but how GANs convert an unstructured latent code to a high quality output is still an open question. In this work, we investiga…
View article: Improving Inversion and Generation Diversity in StyleGAN using a Gaussianized Latent Space
Improving Inversion and Generation Diversity in StyleGAN using a Gaussianized Latent Space Open
Modern Generative Adversarial Networks are capable of creating artificial, photorealistic images from latent vectors living in a low-dimensional learned latent space. It has been shown that a wide range of images can be projected into this…
View article: Seeing What a GAN Cannot Generate
Seeing What a GAN Cannot Generate Open
Despite the success of Generative Adversarial Networks (GANs), mode collapse remains a serious issue during GAN training. To date, little work has focused on understanding and quantifying which modes have been dropped by a model. In this w…
View article: Competitive Collaboration: Joint Unsupervised Learning of Depth, Camera Motion, Optical Flow and Motion Segmentation
Competitive Collaboration: Joint Unsupervised Learning of Depth, Camera Motion, Optical Flow and Motion Segmentation Open
We address the unsupervised learning of several interconnected problems in low-level vision: single view depth prediction, camera motion estimation, optical flow, and segmentation of a video into the static scene and moving regions. Our ke…
View article: Temporal Interpolation as an Unsupervised Pretraining Task for Optical Flow Estimation
Temporal Interpolation as an Unsupervised Pretraining Task for Optical Flow Estimation Open
The difficulty of annotating training data is a major obstacle to using CNNs for low-level tasks in video. Synthetic data often does not generalize to real videos, while unsupervised methods require heuristic losses. Proxy tasks can overco…
View article: Competitive Collaboration: Joint Unsupervised Learning of Depth, Camera\n Motion, Optical Flow and Motion Segmentation
Competitive Collaboration: Joint Unsupervised Learning of Depth, Camera\n Motion, Optical Flow and Motion Segmentation Open
We address the unsupervised learning of several interconnected problems in\nlow-level vision: single view depth prediction, camera motion estimation,\noptical flow, and segmentation of a video into the static scene and moving\nregions. Our…
View article: Model-based Optical Flow: Layers, Learning, and Geometry
Model-based Optical Flow: Layers, Learning, and Geometry Open
The estimation of motion in video sequences establishes temporal correspondences between pixels and surfaces and allows reasoning about a scene using multiple frames. Despite being a focus of research for over three decades, computing moti…
View article: Optical Flow in Mostly Rigid Scenes
Optical Flow in Mostly Rigid Scenes Open
The optical flow of natural scenes is a combination of the motion of the observer and the independent motion of objects. Existing algorithms typically focus on either recovering motion and structure under the assumption of a purely static …