Vlado Menkovski

Very excited to share this! Looking for candidates for a #PhD position on the intersection of #GenerativeAI and Material Science. Supervised by Prof. Sofia Calero and co-supervised by me. jobs.tue.nl/en/vacancy/phd…

An excellent opportunity to work on the intersection of ML and Physics. A vacancy on Machine Learning-based Simulation for Materials Discovery. Supervised by prof. Sofia Calero and co-supervised by me: jobs.tue.nl/en/vacancy/phd…

For more details, we invite you to read the full paper, or come talk to us at ICML 2022! 10/10 🧵

3) LSBD representations also satisfy previous disentanglement notions: Good scores for D_LSBD (lower is better) typically imply good scores on traditional disentanglement metrics (higher is better), even if the reverse is not true. 9/

2) LSBD-VAE and other LSBD methods *can* learn LSBD representations with limited supervision on transformations: Methods that specifically target LSBD indeed score better on D_LSBD. 8/

1) Traditional disentanglement methods don't learn LSBD representations: Methods without a notion of the underlying symmetry group structure don't score well on D_LSBD, even if they do score well on traditional disentanglement metrics. 7/

We perform experiments on a number of datasets with underlying symmetries, using our own LSBD-VAE and other LSBD-oriented methods, as well as various traditional disentanglement methods. From this we draw 3 main conclusions: 6/

We show an intuitive justification of our D_LSBD metric, as well as its theoretical derivation. We also provide a practical implementation to compute D_LSBD for SO(2) symmetry groups of 2D rotations. 5/

In our paper we present the following contributions: i) D_LSBD, a well-formalized general metric to quantify LSBD in learned representations. ii) LSBD-VAE, a weakly-supervised method to learn LSBD representations under certain assumptions. 4/

However, there is currently no metric to quantify LSBD. Such a metric is crucial to evaluate methods aimed at learning LSBD, and to compare to previous notions of disentanglement. 3/

Disentanglement is an important goal in representation learning, but there is no consensus on a formal definition. Higgins et al. (2018) propose such a definition based on the idea that real-world symmetries provide a structure to disentangle, which we refer to as LSBD. 2/

Our paper "Quantifying and Learning Linear Symmetry-Based Disentanglement" got accepted at #ICML2022 @icmlconf!🥳🎉 Congrats to co-authors @tezcatlipocta (joint first), @vlamen, Mike Holenderski & Jim Portegies. Paper: arxiv.org/abs/2011.06070 Code: github.com/luis-armando-p… 1/🧵

The @FullyChargedShw visits Eindhoven and @SolarEHV to check out Stella Vita the solar camper youtu.be/4vHuldVlpFA

Best Paper Award at #icpm21 goes to .... Dominique Sommers, Vlado Menkovski and Dirk Fahland @dfahland with the paper Process Discovery using Graph Neural Networks

We have a vacancy for PhD in Artificial Intelligence for Computational Pathology. Goal: develop new methods for end-to-end training of DL models from whole slide images with application in oncology. #digitalpathology #deeplearning #machinelearning #AI jobs.tue.nl/en/vacancy/phd…

Can Machine Learning be a driver for scientific discovery? Join our multidisciplinary team of scientists and help us push Machine Learning forward to drive progress in natural sciences! jobs.tue.nl/en/vacancy/phd… via @TU/e

Pedestrian orientation dynamics from high-fidelity measurements. Open Access on Scientific Reports nature.com/articles/s4159… with J willems, @vlamen @ftoschi

New preprint: "Pedestrian orientation dynamics from high-fidelity measurements" researchgate.net/publication/33…

Detection of pre-microRNA with Convolutional Neural Networks dlvr.it/RJDLDR

New preprint available 😀 "Deep learning velocity signals allows to quantify turbulence intensity" arxiv.org/abs/1911.05718 with @vlamen, Roberto Benzi and @ftoschi