Search Results for author: Georg Martius

Found 64 papers, 28 papers with code

Colored Noise in PPO: Improved Exploration and Performance Through Correlated Action Sampling

no code implementations18 Dec 2023 Jakob Hollenstein, Georg Martius, Justus Piater

Proximal Policy Optimization (PPO), a popular on-policy deep reinforcement learning method, employs a stochastic policy for exploration.


Regularity as Intrinsic Reward for Free Play

no code implementations NeurIPS 2023 Cansu Sancaktar, Justus Piater, Georg Martius

Our generalized formulation of Regularity as Intrinsic Reward (RaIR) allows us to operationalize it within model-based reinforcement learning.

Model-based Reinforcement Learning reinforcement-learning

Multi-View Causal Representation Learning with Partial Observability

no code implementations7 Nov 2023 Dingling Yao, Danru Xu, Sébastien Lachapelle, Sara Magliacane, Perouz Taslakian, Georg Martius, Julius von Kügelgen, Francesco Locatello

We present a unified framework for studying the identifiability of representations learned from simultaneously observed views, such as different data modalities.

Contrastive Learning Disentanglement

Learning Diverse Skills for Local Navigation under Multi-constraint Optimality

no code implementations3 Oct 2023 Jin Cheng, Marin Vlastelica, Pavel Kolev, Chenhao Li, Georg Martius

We demonstrate the effectiveness of our method on a local navigation task where a quadruped robot needs to reach the target within a finite horizon.

Emergent mechanisms for long timescales depend on training curriculum and affect performance in memory tasks

no code implementations22 Sep 2023 Sina Khajehabdollahi, Roxana Zeraati, Emmanouil Giannakakis, Tim Jakob Schäfer, Georg Martius, Anna Levina

We find that for both tasks RNNs develop longer timescales with increasing $N$, but depending on the learning objective, they use different mechanisms.

Mind the Uncertainty: Risk-Aware and Actively Exploring Model-Based Reinforcement Learning

no code implementations11 Sep 2023 Marin Vlastelica, Sebastian Blaes, Cristina Pineri, Georg Martius

We introduce a simple but effective method for managing risk in model-based reinforcement learning with trajectory sampling that involves probabilistic safety constraints and balancing of optimism in the face of epistemic uncertainty and pessimism in the face of aleatoric uncertainty of an ensemble of stochastic neural networks. Various experiments indicate that the separation of uncertainties is essential to performing well with data-driven MPC approaches in uncertain and safety-critical control environments.

Model-based Reinforcement Learning reinforcement-learning

Benchmarking Offline Reinforcement Learning on Real-Robot Hardware

2 code implementations28 Jul 2023 Nico Gürtler, Sebastian Blaes, Pavel Kolev, Felix Widmaier, Manuel Wüthrich, Stefan Bauer, Bernhard Schölkopf, Georg Martius

To coordinate the efforts of the research community toward tackling this problem, we propose a benchmark including: i) a large collection of data for offline learning from a dexterous manipulation platform on two tasks, obtained with capable RL agents trained in simulation; ii) the option to execute learned policies on a real-world robotic system and a simulation for efficient debugging.

Benchmarking reinforcement-learning

Diverse Offline Imitation Learning

no code implementations21 Jul 2023 Marin Vlastelica, Jin Cheng, Georg Martius, Pavel Kolev

There has been significant recent progress in the area of unsupervised skill discovery, utilizing various information-theoretic objectives as measures of diversity.

D4RL Imitation Learning

The Expressive Leaky Memory Neuron: an Efficient and Expressive Phenomenological Neuron Model Can Solve Long-Horizon Tasks

1 code implementation14 Jun 2023 Aaron Spieler, Nasim Rahaman, Georg Martius, Bernhard Schölkopf, Anna Levina

Biological cortical neurons are remarkably sophisticated computational devices, temporally integrating their vast synaptic input over an intricate dendritic tree, subject to complex, nonlinearly interacting internal biological processes.

Classification Long-range modeling +3

Locally adaptive cellular automata for goal-oriented self-organization

no code implementations12 Jun 2023 Sina Khajehabdollahi, Emmanouil Giannakakis, Victor Buendia, Georg Martius, Anna Levina

In this paper, we propose a new model class of adaptive cellular automata that allows for the generation of scalable and expressive models.

Object-Centric Learning for Real-World Videos by Predicting Temporal Feature Similarities

1 code implementation NeurIPS 2023 Andrii Zadaianchuk, Maximilian Seitzer, Georg Martius

Recently, it was shown that the reconstruction of pre-trained self-supervised features leads to object-centric representations on unconstrained real-world image datasets.

Object Object Discovery

Discovering Causal Relations and Equations from Data

no code implementations21 May 2023 Gustau Camps-Valls, Andreas Gerhardus, Urmi Ninad, Gherardo Varando, Georg Martius, Emili Balaguer-Ballester, Ricardo Vinuesa, Emiliano Diaz, Laure Zanna, Jakob Runge

Discovering equations, laws and principles that are invariant, robust and causal explanations of the world has been fundamental in physical sciences throughout the centuries.


When to be critical? Performance and evolvability in different regimes of neural Ising agents

no code implementations28 Mar 2023 Sina Khajehabdollahi, Jan Prosi, Emmanouil Giannakakis, Georg Martius, Anna Levina

To test it we introduce a hard and simple task: for the hard task, agents evolve closer to criticality whereas more subcritical solutions are found for the simple task.

Efficient Learning of High Level Plans from Play

no code implementations16 Mar 2023 Núria Armengol Urpí, Marco Bagatella, Otmar Hilliges, Georg Martius, Stelian Coros

Real-world robotic manipulation tasks remain an elusive challenge, since they involve both fine-grained environment interaction, as well as the ability to plan for long-horizon goals.

Motion Planning Reinforcement Learning (RL) +1

Learning with Muscles: Benefits for Data-Efficiency and Robustness in Anthropomorphic Tasks

no code implementations8 Jul 2022 Isabell Wochner, Pierre Schumacher, Georg Martius, Dieter Büchler, Syn Schmitt, Daniel F. B. Haeufle

Humans are able to outperform robots in terms of robustness, versatility, and learning of new tasks in a wide variety of movements.

Curious Exploration via Structured World Models Yields Zero-Shot Object Manipulation

no code implementations22 Jun 2022 Cansu Sancaktar, Sebastian Blaes, Georg Martius

It has been a long-standing dream to design artificial agents that explore their environment efficiently via intrinsic motivation, similar to how children perform curious free play.

Efficient Exploration Object +2

Backpropagation through Combinatorial Algorithms: Identity with Projection Works

2 code implementations30 May 2022 Subham Sekhar Sahoo, Anselm Paulus, Marin Vlastelica, Vít Musil, Volodymyr Kuleshov, Georg Martius

Embedding discrete solvers as differentiable layers has given modern deep learning architectures combinatorial expressivity and discrete reasoning capabilities.

Density Estimation Graph Matching +3

On the Pitfalls of Heteroscedastic Uncertainty Estimation with Probabilistic Neural Networks

2 code implementations ICLR 2022 Maximilian Seitzer, Arash Tavakoli, Dimitrije Antic, Georg Martius

In this work, we examine this approach and identify potential hazards associated with the use of log-likelihood in conjunction with gradient-based optimizers.

Hierarchical Reinforcement Learning with Timed Subgoals

1 code implementation NeurIPS 2021 Nico Gürtler, Dieter Büchler, Georg Martius

Hierarchical reinforcement learning (HRL) holds great potential for sample-efficient learning on challenging long-horizon tasks.

Hierarchical Reinforcement Learning reinforcement-learning +1

A soft thumb-sized vision-based sensor with accurate all-round force perception

no code implementations10 Nov 2021 Huanbo Sun, Katherine J. Kuchenbecker, Georg Martius

Insight has an overall spatial resolution of 0. 4 mm, force magnitude accuracy around 0. 03 N, and force direction accuracy around 5 degrees over a range of 0. 03--2 N for numerous distinct contacts with varying contact area.

Sparsely Changing Latent States for Prediction and Planning in Partially Observable Domains

1 code implementation NeurIPS 2021 Christian Gumbsch, Martin V. Butz, Georg Martius

A common approach to prediction and planning in partially observable domains is to use recurrent neural networks (RNNs), which ideally develop and maintain a latent memory about hidden, task-relevant factors.

Inductive Bias

Planning from Pixels in Environments with Combinatorially Hard Search Spaces

2 code implementations NeurIPS 2021 Marco Bagatella, Mirek Olšák, Michal Rolínek, Georg Martius

The ability to form complex plans based on raw visual input is a litmus test for current capabilities of artificial intelligence, as it requires a seamless combination of visual processing and abstract algorithmic execution, two traditionally separate areas of computer science.

Continuous Control Offline RL

Self-supervised Reinforcement Learning with Independently Controllable Subgoals

no code implementations9 Sep 2021 Andrii Zadaianchuk, Georg Martius, Fanny Yang

We propose a novel self-supervised agent that estimates relations between environment components and uses them to independently control different parts of the environment state.

reinforcement-learning Reinforcement Learning (RL)

Assessing aesthetics of generated abstract images using correlation structure

no code implementations18 May 2021 Sina Khajehabdollahi, Georg Martius, Anna Levina

We demonstrate that even with the randomly selected weights the correlation functions remain largely determined by the network architecture.

Informed Equation Learning

no code implementations13 May 2021 Matthias Werner, Andrej Junginger, Philipp Hennig, Georg Martius

Our system then utilizes a robust method to learn equations with atomic functions exhibiting singularities, as e. g. logarithm and division.

CombOptNet: Fit the Right NP-Hard Problem by Learning Integer Programming Constraints

1 code implementation5 May 2021 Anselm Paulus, Michal Rolínek, Vít Musil, Brandon Amos, Georg Martius

Bridging logical and algorithmic reasoning with modern machine learning techniques is a fundamental challenge with potentially transformative impact.

The dynamical regime and its importance for evolvability, task performance and generalization

2 code implementations22 Mar 2021 Jan Prosi, Sina Khajehabdollahi, Emmanouil Giannakakis, Georg Martius, Anna Levina

Surprisingly, we find that all populations, regardless of their initial regime, evolve to be subcritical in simple tasks and even strongly subcritical populations can reach comparable performance.

Neuro-algorithmic Policies enable Fast Combinatorial Generalization

no code implementations15 Feb 2021 Marin Vlastelica, Michal Rolínek, Georg Martius

Furthermore, we show that for a certain subclass of the MDP framework, this can be alleviated by neuro-algorithmic architectures.

Demystifying Inductive Biases for $β$-VAE Based Architectures

no code implementations12 Feb 2021 Dominik Zietlow, Michal Rolinek, Georg Martius

By small, elaborate perturbations of existing datasets, we hide the convenient correlation structure that is easily exploited by a variety of architectures.

Disentanglement Inductive Bias

Machine learning time-local generators of open quantum dynamics

no code implementations21 Jan 2021 Paolo P. Mazza, Dominik Zietlow, Federico Carollo, Sabine Andergassen, Georg Martius, Igor Lesanovsky

Such evolution is typically emerging under the assumption of a weak coupling between the system and an infinitely large bath.

Quantum Physics Quantum Gases

Neuro-algorithmic Policies for Discrete Planning

no code implementations1 Jan 2021 Marin Vlastelica Pogančić, Michal Rolinek, Georg Martius

Although model-based and model-free approaches to learning control of systems have achieved impressive results on standard benchmarks, most have been shown to be lacking in their generalization capabilities.

Clearing the Path for Truly Semantic Representation Learning

no code implementations1 Jan 2021 Dominik Zietlow, Michal Rolinek, Georg Martius

The performance of $\beta$-Variational-Autoencoders ($\beta$-VAEs) and their variants on learning semantically meaningful, disentangled representations is unparalleled.


How to Train Your Differentiable Filter

1 code implementation28 Dec 2020 Alina Kloss, Georg Martius, Jeannette Bohg

In many robotic applications, it is crucial to maintain a belief about the state of a system, which serves as input for planning and decision making and provides feedback during task execution.

Decision Making

Self-supervised Visual Reinforcement Learning with Object-centric Representations

1 code implementation ICLR 2021 Andrii Zadaianchuk, Maximilian Seitzer, Georg Martius

We show that the structure in the representations in combination with goal-conditioned attention policies helps the autonomous agent to discover and learn useful skills.

Object reinforcement-learning +1

Discrete Planning with Neuro-algorithmic Policies

no code implementations NeurIPS Workshop LMCA 2020 Marin Vlastelica Pogančić, Michal Rolinek, Georg Martius

Although model-based and model-free approaches to learning the control of systems have achieved impressive results on standard benchmarks, most have been shown to be lacking in their generalization capabilities.

Fit The Right NP-Hard Problem: End-to-end Learning of Integer Programming Constraints

no code implementations NeurIPS Workshop LMCA 2020 Anselm Paulus, Michal Rolinek, Vít Musil, Brandon Amos, Georg Martius

Bridging logical and algorithmic reasoning with modern machine learning techniques is a fundamental challenge with potentially transformative impact.

Deep Graph Matching via Blackbox Differentiation of Combinatorial Solvers

5 code implementations25 Mar 2020 Michal Rolínek, Paul Swoboda, Dominik Zietlow, Anselm Paulus, Vít Musil, Georg Martius

Building on recent progress at the intersection of combinatorial optimization and deep learning, we propose an end-to-end trainable architecture for deep graph matching that contains unmodified combinatorial solvers.

Combinatorial Optimization Graph Matching

Optimizing Rank-based Metrics with Blackbox Differentiation

1 code implementation7 Dec 2019 Michal Rolínek, Vít Musil, Anselm Paulus, Marin Vlastelica, Claudio Michaelis, Georg Martius

Rank-based metrics are some of the most widely used criteria for performance evaluation of computer vision models.

Image Retrieval object-detection +2

Differentiation of Blackbox Combinatorial Solvers

6 code implementations ICLR 2020 Marin Vlastelica, Anselm Paulus, Vít Musil, Georg Martius, Michal Rolínek

Achieving fusion of deep learning with combinatorial algorithms promises transformative changes to artificial intelligence.

Traveling Salesman Problem

Fast Non-Parametric Learning to Accelerate Mixed-Integer Programming for Online Hybrid Model Predictive Control

no code implementations20 Nov 2019 Jia-Jie Zhu, Georg Martius

Today's fast linear algebra and numerical optimization tools have pushed the frontier of model predictive control (MPC) forward, to the efficient control of highly nonlinear and hybrid systems.

Model Predictive Control

Analytical classical density functionals from an equation learning network

1 code implementation28 Oct 2019 Shang-Chun Lin, Georg Martius, Martin Oettel

We explore the feasibility of using machine learning methods to obtain an analytic form of the classical free energy functional for two model fluids, hard rods and Lennard--Jones, in one dimension .

BIG-bench Machine Learning

Control What You Can: Intrinsically Motivated Task-Planning Agent

1 code implementation NeurIPS 2019 Sebastian Blaes, Marin Vlastelica Pogančić, Jia-Jie Zhu, Georg Martius

We present a novel intrinsically motivated agent that learns how to control the environment in the fastest possible manner by optimizing learning progress.

Autonomous Identification and Goal-Directed Invocation of Event-Predictive Behavioral Primitives

no code implementations26 Feb 2019 Christian Gumbsch, Martin V. Butz, Georg Martius

Here, we introduce a computational learning architecture, termed surprise-based behavioral modularization into event-predictive structures (SUBMODES), that explores behavior and identifies the underlying behavioral units completely from scratch.

Variational Autoencoders Pursue PCA Directions (by Accident)

no code implementations CVPR 2019 Michal Rolinek, Dominik Zietlow, Georg Martius

The Variational Autoencoder (VAE) is a powerful architecture capable of representation learning and generative modeling.

Representation Learning

Deep Reinforcement Learning for Event-Triggered Control

1 code implementation13 Sep 2018 Dominik Baumann, Jia-Jie Zhu, Georg Martius, Sebastian Trimpe

Event-triggered control (ETC) methods can achieve high-performance control with a significantly lower number of samples compared to usual, time-triggered methods.

reinforcement-learning Reinforcement Learning (RL)

Extrapolation and learning equations

1 code implementation10 Oct 2016 Georg Martius, Christoph H. Lampert

In classical machine learning, regression is treated as a black box process of identifying a suitable function from a hypothesis set without attempting to gain insight into the mechanism connecting inputs and outputs.

Self-organized control for musculoskeletal robots

1 code implementation9 Feb 2016 Ralf Der, Georg Martius

The paper presents a solution with a controller that is devoid of any functionalities of its own, given by a fixed, explicit and context-free function of the recent history of the sensor values.

Quantifying Emergent Behavior of Autonomous Robots

1 code implementation6 Oct 2015 Georg Martius, Eckehard Olbrich

For deterministic systems both measures will diverge with increasing resolution.

Time Series Time Series Analysis

A novel plasticity rule can explain the development of sensorimotor intelligence

no code implementations4 May 2015 Ralf Der, Georg Martius

Grounding autonomous behavior in the nervous system is a fundamental challenge for neuroscience.

Self-Learning Test

Linear combination of one-step predictive information with an external reward in an episodic policy gradient setting: a critical analysis

no code implementations26 Sep 2013 Keyan Zahedi, Georg Martius, Nihat Ay

Previous experiments have shown that the predictive information (PI) is a good candidate to support autonomous, open-ended learning of complex behaviours, because a maximisation of the PI corresponds to an exploration of morphology- and environment-dependent behavioural regularities.

Information driven self-organization of complex robotic behaviors

no code implementations30 Jan 2013 Georg Martius, Ralf Der, Nihat Ay

We study nonlinear and nonstationary systems and introduce the time-local predicting information (TiPI) which allows us to derive exact results together with explicit update rules for the parameters of the controller in the dynamical systems framework.

Cannot find the paper you are looking for? You can Submit a new open access paper.