Starting from Sunday the 2nd of December, the beautiful city of Montréal, Canada, will host one of the most exciting and famous AI conference in the world.
NeurIPS, or Neural Information Processing Systems, is one of the biggest event where every scientist from undergraduate to chair of departments aim to go.
I am very proud and honored to be there this year at NeurIPS 2018 to present sc-SchNet, my work on interpratibility of Deep Neural Networks for atomistic systems. This work has been done in tight collaboration with my postdocs colleagues at the ML department at TU in Berlin. Here’s the abstract:
In this work, we extend the SchNet architecture by using weighted skip connections to assemble the final representation. This enables us to study the relative importance of each interaction block for property prediction. We demonstrate on both the QM9 and MD17 dataset that their relative weighting depends strongly on the chemical composition and configurational degrees of freedom of the molecules which opens the path towards a more detailed understanding of machine learning models for molecules.
If you are intersted and want to know more you can find the paper here. I will be presenting a poster during the morning poster session at the 2nd Workshop on Molecules and Materials on Saturday the 8th. The venue of the conference is the Palais des Congrés in Montréal. If you are planning to come, please feel free to swing by with any questions. I will be happy to talk about it and meet new folks working in the field!
See you in Canada! Cheers,