The new method outperforms baselines from topological change detection and anomaly detection by a large margin and competes with supervised models on one of the datasets. Code available at https://t.co/R7the0mQzj 5/5

We introduce the first dataset with annotated topological changes to aid evaluation in this field, which has been relying on qualitative assessment of small sample sizes and proxy-tasks before. 4/5

The method is based on probabilistic registration models, using a VAE to give pixel-wise bounds on the likelihood of a diffeomorphic match. We extend prior work by re-defining the output distribution and eliminating hyperparameters in the prior. 3/5

TLDR: Alignment of distributions or images often assumes a common topology, with undefined behaviour if this assumption is violated. Our method detects+marks violations of this assumption. Can inform topology-aware registration models or warn about areas of wrong alignment. 2/5

Can we detect when geometric alignment algorithms make errors? 🔍 We introduce the first unsupervised, learning-based method for detecting changes in image topology, along with a new dataset for evaluation. Accepted to #neurips2021 https://t.co/0Tcn1ZrN3b 1/5

A very interesting take at machine learning as "Software 2.0", it's pros and cons compared to traditional "Software 1.0" https://t.co/ejQQ0eUIDP

Want to understand the LDDMM diffeomorphic image registration algorithm by @drmfbeg? I published a simple, fully commented python implementation. The code references equations of the original paper and makes learning simple. https://t.co/AwWQowAbsk