4 papers
Blind Recovery of Latent Domains via Unsupervised Symmetry Discovery
Onur Efe, Arkadas Ozakin
Primary motivation in blind inverse problems is to recover signals of interest from corrupted observations without knowing the obfuscating mechanism. Blind deconvolution is a promi…
Learning Relativistic Geodesics and Chaotic Dynamics via Stabilized Lagrangian Neural Networks
Abdullah Umut Hamzaogullari, Arkadas Ozakin
Lagrangian Neural Networks (LNNs) can learn arbitrary Lagrangians from trajectory data, but their unusual optimization objective leads to significant training instabilities that li…
SymmetryLens: Unsupervised Symmetry Learning via Locality and Density Preservation
Onur Efe, Arkadas Ozakin
We develop a new unsupervised symmetry learning method that starts with raw data and provides the minimal generator of an underlying Lie group of symmetries, together with a symmet…
RECOVAR: Representation Covariances on Deep Latent Spaces for Seismic Event Detection
Onur Efe, Arkadas Ozakin
While modern deep learning methods have shown great promise in the problem of earthquake detection, the most successful methods so far have been based on supervised learning, which…