Search Results for author: Mathias Lechner

Found 41 papers, 22 papers with code

The Natural Lottery Ticket Winner: Reinforcement Learning with Ordinary Neural Circuits

no code implementations ICML 2020 Ramin Hasani, Mathias Lechner, Alexander Amini, Daniela Rus, Radu Grosu

We propose a neural information processing system which is obtained by re-purposing the function of a biological neural circuit model to govern simulated and real-world control tasks.

reinforcement-learning Reinforcement Learning (RL)

Compositional Policy Learning in Stochastic Control Systems with Formal Guarantees

1 code implementation NeurIPS 2023 Đorđe Žikelić, Mathias Lechner, Abhinav Verma, Krishnendu Chatterjee, Thomas A. Henzinger

We also derive a tighter lower bound compared to previous work on the probability of reach-avoidance implied by a RASM, which is required to find a compositional policy with an acceptable probabilistic threshold for complex tasks with multiple edge policies.

Learning with Chemical versus Electrical Synapses -- Does it Make a Difference?

no code implementations21 Nov 2023 Mónika Farsang, Mathias Lechner, David Lung, Ramin Hasani, Daniela Rus, Radu Grosu

In this work we aim to determine the impact of using chemical synapses compared to electrical synapses, in both sparse and all-to-all connected networks.

Autonomous Driving

Leveraging Low-Rank and Sparse Recurrent Connectivity for Robust Closed-Loop Control

no code implementations5 Oct 2023 Neehal Tumma, Mathias Lechner, Noel Loo, Ramin Hasani, Daniela Rus

In this work, we explore the application of recurrent neural networks to tasks of this nature and understand how a parameterization of their recurrent connectivity influences robustness in closed-loop settings.

On the Size and Approximation Error of Distilled Sets

no code implementations23 May 2023 Alaa Maalouf, Murad Tukan, Noel Loo, Ramin Hasani, Mathias Lechner, Daniela Rus

Despite significant empirical progress in recent years, there is little understanding of the theoretical limitations/guarantees of dataset distillation, specifically, what excess risk is achieved by distillation compared to the original dataset, and how large are distilled datasets?

regression

Infrastructure-based End-to-End Learning and Prevention of Driver Failure

no code implementations21 Mar 2023 Noam Buckman, Shiva Sreeram, Mathias Lechner, Yutong Ban, Ramin Hasani, Sertac Karaman, Daniela Rus

FailureNet observes the poses of vehicles as they approach an intersection and detects whether a failure is present in the autonomy stack, warning cross-traffic of potentially dangerous drivers.

Autonomous Vehicles

Dataset Distillation with Convexified Implicit Gradients

2 code implementations13 Feb 2023 Noel Loo, Ramin Hasani, Mathias Lechner, Daniela Rus

We propose a new dataset distillation algorithm using reparameterization and convexification of implicit gradients (RCIG), that substantially improves the state-of-the-art.

Understanding Reconstruction Attacks with the Neural Tangent Kernel and Dataset Distillation

1 code implementation2 Feb 2023 Noel Loo, Ramin Hasani, Mathias Lechner, Alexander Amini, Daniela Rus

We show that both theoretically and empirically, reconstructed images tend to "outliers" in the dataset, and that these reconstruction attacks can be used for \textit{dataset distillation}, that is, we can retrain on reconstructed images and obtain high predictive accuracy.

Reconstruction Attack

Towards Cooperative Flight Control Using Visual-Attention

no code implementations21 Dec 2022 Lianhao Yin, Makram Chahine, Tsun-Hsuan Wang, Tim Seyde, Chao Liu, Mathias Lechner, Ramin Hasani, Daniela Rus

We propose an air-guardian system that facilitates cooperation between a pilot with eye tracking and a parallel end-to-end neural control system.

Feature Importance

Learning Control Policies for Stochastic Systems with Reach-avoid Guarantees

no code implementations11 Oct 2022 Đorđe Žikelić, Mathias Lechner, Thomas A. Henzinger, Krishnendu Chatterjee

We study the problem of learning controllers for discrete-time non-linear stochastic dynamical systems with formal reach-avoid guarantees.

Learning Provably Stabilizing Neural Controllers for Discrete-Time Stochastic Systems

1 code implementation11 Oct 2022 Matin Ansaripour, Krishnendu Chatterjee, Thomas A. Henzinger, Mathias Lechner, Đorđe Žikelić

We show that this procedure can also be adapted to formally verifying that, under a given Lipschitz continuous control policy, the stochastic system stabilizes within some stabilizing region with probability~$1$.

Continuous Control

On the Forward Invariance of Neural ODEs

no code implementations10 Oct 2022 Wei Xiao, Tsun-Hsuan Wang, Ramin Hasani, Mathias Lechner, Yutong Ban, Chuang Gan, Daniela Rus

We propose a new method to ensure neural ordinary differential equations (ODEs) satisfy output specifications by using invariance set propagation.

Autonomous Vehicles Collision Avoidance +2

Are All Vision Models Created Equal? A Study of the Open-Loop to Closed-Loop Causality Gap

no code implementations9 Oct 2022 Mathias Lechner, Ramin Hasani, Alexander Amini, Tsun-Hsuan Wang, Thomas A. Henzinger, Daniela Rus

Our results imply that the causality gap can be solved in situation one with our proposed training guideline with any modern network architecture, whereas achieving out-of-distribution generalization (situation two) requires further investigations, for instance, on data diversity rather than the model architecture.

Autonomous Driving Image Classification +1

Liquid Structural State-Space Models

1 code implementation26 Sep 2022 Ramin Hasani, Mathias Lechner, Tsun-Hsuan Wang, Makram Chahine, Alexander Amini, Daniela Rus

A proper parametrization of state transition matrices of linear state-space models (SSMs) followed by standard nonlinearities enables them to efficiently learn representations from sequential data, establishing the state-of-the-art on a large series of long-range sequence modeling benchmarks.

Heart rate estimation Long-range modeling +3

Entangled Residual Mappings

no code implementations2 Jun 2022 Mathias Lechner, Ramin Hasani, Zahra Babaiee, Radu Grosu, Daniela Rus, Thomas A. Henzinger, Sepp Hochreiter

Residual mappings have been shown to perform representation learning in the first layers and iterative feature refinement in higher layers.

Inductive Bias Representation Learning

Learning Stabilizing Policies in Stochastic Control Systems

no code implementations24 May 2022 Đorđe Žikelić, Mathias Lechner, Krishnendu Chatterjee, Thomas A. Henzinger

In this work, we address the problem of learning provably stable neural network policies for stochastic control systems.

Revisiting the Adversarial Robustness-Accuracy Tradeoff in Robot Learning

no code implementations15 Apr 2022 Mathias Lechner, Alexander Amini, Daniela Rus, Thomas A. Henzinger

However, the improved robustness does not come for free but rather is accompanied by a decrease in overall model accuracy and performance.

Adversarial Robustness Autonomous Driving +2

Stability Verification in Stochastic Control Systems via Neural Network Supermartingales

no code implementations17 Dec 2021 Mathias Lechner, Đorđe Žikelić, Krishnendu Chatterjee, Thomas A. Henzinger

We consider the problem of formally verifying almost-sure (a. s.) asymptotic stability in discrete-time nonlinear stochastic control systems.

Infinite Time Horizon Safety of Bayesian Neural Networks

1 code implementation NeurIPS 2021 Mathias Lechner, Đorđe Žikelić, Krishnendu Chatterjee, Thomas A. Henzinger

Bayesian neural networks (BNNs) place distributions over the weights of a neural network to model uncertainty in the data and the network's prediction.

reinforcement-learning Reinforcement Learning (RL) +1

Interactive Analysis of CNN Robustness

1 code implementation14 Oct 2021 Stefan Sietzen, Mathias Lechner, Judy Borowski, Ramin Hasani, Manuela Waldner

While convolutional neural networks (CNNs) have found wide adoption as state-of-the-art models for image-related tasks, their predictions are often highly sensitive to small input perturbations, which the human vision is robust against.

GoTube: Scalable Stochastic Verification of Continuous-Depth Models

1 code implementation18 Jul 2021 Sophie Gruenbacher, Mathias Lechner, Ramin Hasani, Daniela Rus, Thomas A. Henzinger, Scott Smolka, Radu Grosu

Our algorithm solves a set of global optimization (Go) problems over a given time horizon to construct a tight enclosure (Tube) of the set of all process executions starting from a ball of initial states.

Closed-form Continuous-time Neural Models

1 code implementation25 Jun 2021 Ramin Hasani, Mathias Lechner, Alexander Amini, Lucas Liebenwein, Aaron Ray, Max Tschaikowski, Gerald Teschl, Daniela Rus

To this end, we compute a tightly-bounded approximation of the solution of an integral appearing in LTCs' dynamics, that has had no known closed-form solution so far.

Sentiment Analysis Time Series Prediction

Causal Navigation by Continuous-time Neural Networks

1 code implementation NeurIPS 2021 Charles Vorbach, Ramin Hasani, Alexander Amini, Mathias Lechner, Daniela Rus

We evaluate our method in the context of visual-control learning of drones over a series of complex tasks, ranging from short- and long-term navigation, to chasing static and dynamic objects through photorealistic environments.

Imitation Learning

Adversarial Training is Not Ready for Robot Learning

no code implementations15 Mar 2021 Mathias Lechner, Ramin Hasani, Radu Grosu, Daniela Rus, Thomas A. Henzinger

Adversarial training is an effective method to train deep learning models that are resilient to norm-bounded perturbations, with the cost of nominal performance drop.

Latent Imagination Facilitates Zero-Shot Transfer in Autonomous Racing

1 code implementation8 Mar 2021 Axel Brunnbauer, Luigi Berducci, Andreas Brandstätter, Mathias Lechner, Ramin Hasani, Daniela Rus, Radu Grosu

World models learn behaviors in a latent imagination space to enhance the sample-efficiency of deep reinforcement learning (RL) algorithms.

Continuous Control Reinforcement Learning (RL)

On The Verification of Neural ODEs with Stochastic Guarantees

no code implementations16 Dec 2020 Sophie Gruenbacher, Ramin Hasani, Mathias Lechner, Jacek Cyranka, Scott A. Smolka, Radu Grosu

We show that Neural ODEs, an emerging class of time-continuous neural networks, can be verified by solving a set of global-optimization problems.

Scalable Verification of Quantized Neural Networks (Technical Report)

1 code implementation15 Dec 2020 Thomas A. Henzinger, Mathias Lechner, Đorđe Žikelić

In this paper, we show that verifying the bit-exact implementation of quantized neural networks with bit-vector specifications is PSPACE-hard, even though verifying idealized real-valued networks and satisfiability of bit-vector specifications alone are each in NP.

Computational Efficiency Quantization

Lagrangian Reachtubes: The Next Generation

1 code implementation14 Dec 2020 Sophie Gruenbacher, Jacek Cyranka, Mathias Lechner, Md. Ariful Islam, Scott A. Smolka, Radu Grosu

Secondly, it computes the next reachset as the intersection of two balls: one based on the Cartesian metric and the other on the new metric.

Neural circuit policies enabling auditable autonomy

1 code implementation13 Oct 2020 Mathias Lechner, Ramin Hasani, Alexander Amini, Thomas A. Henzinger, Daniela Rus & Radu Grosu

A central goal of artificial intelligence in high-stakes decision-making applications is to design a single algorithm that simultaneously expresses generalizability by learning coherent representations of their world and interpretable explanations of its dynamics.

Autonomous Vehicles Decision Making

Learning representations for binary-classification without backpropagation

1 code implementation ICLR 2020 Mathias Lechner

The family of feedback alignment (FA) algorithms aims to provide a more biologically motivated alternative to backpropagation (BP), by substituting the computations that are unrealistic to be implemented in physical brains.

Binary Classification Classification +1

Liquid Time-constant Recurrent Neural Networks as Universal Approximators

no code implementations1 Nov 2018 Ramin M. Hasani, Mathias Lechner, Alexander Amini, Daniela Rus, Radu Grosu

In this paper, we introduce the notion of liquid time-constant (LTC) recurrent neural networks (RNN)s, a subclass of continuous-time RNNs, with varying neuronal time-constant realized by their nonlinear synaptic transmission model.

Can a Compact Neuronal Circuit Policy be Re-purposed to Learn Simple Robotic Control?

1 code implementation11 Sep 2018 Ramin Hasani, Mathias Lechner, Alexander Amini, Daniela Rus, Radu Grosu

Inspired by the structure of the nervous system of the soil-worm, C. elegans, we introduce Neuronal Circuit Policies (NCPs), defined as the model of biological neural circuits reparameterized for the control of an alternative task.

Response Characterization for Auditing Cell Dynamics in Long Short-term Memory Networks

no code implementations11 Sep 2018 Ramin M. Hasani, Alexander Amini, Mathias Lechner, Felix Naser, Radu Grosu, Daniela Rus

In this paper, we introduce a novel method to interpret recurrent neural networks (RNNs), particularly long short-term memory networks (LSTMs) at the cellular level.

Neuronal Circuit Policies

1 code implementation22 Mar 2018 Mathias Lechner, Ramin M. Hasani, Radu Grosu

We propose an effective way to create interpretable control agents, by re-purposing the function of a biological neural circuit model, to govern simulated and real world reinforcement learning (RL) test-beds.

Reinforcement Learning (RL)

Worm-level Control through Search-based Reinforcement Learning

no code implementations9 Nov 2017 Mathias Lechner, Radu Grosu, Ramin M. Hasani

We model the tap-withdrawal (TW) neural circuit of the nematode, \textit{C. elegans}, a circuit responsible for the worm's reflexive response to external mechanical touch stimulations, and learn its synaptic and neural parameters as a policy for controlling the inverted pendulum problem.

reinforcement-learning Reinforcement Learning (RL)

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