Header logo is

Group News

CVPR Oral Now Online

  • 15 June 2020

We just uploaded our presentations for our CVPR paper "Optimizing Rank-based Metrics with Blackbox Differentiation" to YouTube.

Michal Rolinek Anselm Paulus Marin Vlastelica Pogancic Georg Martius


MPI for Intelligent Systems continues video series

  • 29 April 2020

Video No 4 is now available

Short films present scientists' research projects in an understandable way

Alejandro Posada Linda Behringer Georg Martius Alexander Badri-Sprowitz Andreas Geiger Sebastian Trimpe


Accepted Paper at CVPR

  • 01 February 2020

Our paper on optimizing rank-based metrics with blackbox differentiation received flawless reviews and got accepted at CVPR 2020

Michal Rolinek Anselm Paulus Marin Vlastelica Pogancic Georg Martius


Perfect Review-Scores for our ICLR Submission

  • 20 December 2019

Our paper "Differentiation of Blackbox Combinatorial Solvers" received perfect review-scores for ICLR 2020!

Marin Vlastelica Pogancic Anselm Paulus Georg Martius Michal Rolinek


Two posters at NeurIPS

  • 12 December 2019

We presented two posters at NeurIPS: "Control What You Can" and "Differentiation of Blackbox Combinatorial Solvers"

Sebastian Blaes Marin Vlastelica Pogancic Michal Rolinek Georg Martius


Article published in Frontiers in Neurorobotics

  • 10 July 2019

Huanbo's paper 'Machine Learning for Haptics: Inferring Multi-Contact Stimulation From Sparse Sensor Configuration' was successfully published!

Huanbo Sun Georg Martius


CVPR 2019 Paper Accepted

  • 12 March 2019

Our paper 'Variational Autoencoders Recover PCA Directions (by Accident)' got accepted at CVPR 2019! We are looking forward to present it as a poster in June at Long Beach!

The paper elaborates on the close connection between Variational Autoencoders and the well known PCA algorithm in terms of their alignment of the latent space.

Michal Rolinek Dominik Zietlow Georg Martius


Winners of the final RL Course Competition

  • 11 February 2019

Congratulations to Team SSP for winning the final Reinforcement Learning course competition with their SAC (Soft Actor Critic) agent.

Sebastian Blaes Jia-Jie Zhu Georg Martius


Poster @ NeurIPS 2018

  • 04 December 2018

presentation of our L4 Paper

We had a busy 2h poster presentation with many interested visitors


First machine learning method capable of accurate extrapolation

  • 13 July 2018

Scientists develop new machine learning method that can make robots safer - New method provides simpler and more intuitive models of physical situations

Understanding how a robot will react under different conditions is essential to guaranteeing its safe operation. But how do you know what will break a robot without actually damaging it? A new method developed by scientists at the Institute of Science and Technology Austria and the Max Planck Institute for Intelligent Systems is the first machine learning method that can use observations made under safe conditions to make accurate predictions for all possible conditions governed by the same physical dynamics. Especially designed for real-life situations, their method provides simple, interpretable descriptions of the underlying physics. The researchers will present their findings tomorrow at this year’s prestigious International Conference for Machine Learning (ICML).

Georg Martius