A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2022; you can also visit the original URL.
The file type is application/pdf
.
Filters
SABLAS: Learning Safe Control for Black-box Dynamical Systems
2022
IEEE Robotics and Automation Letters
In this paper, we propose a novel method that can learn safe control policies and barrier certificates for black-box dynamical systems, without requiring for an accurate system model. ...
the black-box system. ...
Definition 2 (Safe Control of Black-box Systems): Given a black-box dynamical system modeled as in Definition 1, the safe control problem aims to find a controller π : S → U such that under control input ...
doi:10.1109/lra.2022.3142743
fatcat:hj2hdo46xzeujkuzqmwtc7xlke
Vehicle Detection Using Deep Learning Technique in Tunnel Road Environments
2020
Symmetry
The vehicle detection rate is approximately 87%, while the detection accuracy is approximately 94% for the proposed method applied to various tunnel road environments. ...
This paper proposes a real-time detection method for a car driving ahead in real time on a tunnel road. ...
If the black-box product is equipped with a function for safe driving support, it will be possible to support safe driving. ...
doi:10.3390/sym12122012
fatcat:juehdorkungdreg23ptnvhhm3m
Model-free Safe Control for Zero-Violation Reinforcement Learning
2021
Conference on Robot Learning
In particular, we present an implicit safe set algorithm, which synthesizes the safety index (also called the barrier certificate) and the subsequent safe control law only by querying a black-box dynamic ...
This paper presents a model-free safe control strategy to synthesize safeguards for DRL agents, which will ensure zero safety violation during training. ...
However, for black-box dynamics, this constraint is challenging to quantify. ...
dblp:conf/corl/ZhaoHL21
fatcat:dwgaj47onfcjxcbat3fvdppmvq
Asynchronous Learning for Service Composition
[chapter]
2012
Lecture Notes in Computer Science
. ⇒ learned models only approximate real behaviour ⇒ can we safely build a system using imprecise models?
Input-enabled inference The star: L* (Angluin) algorithm. ...
Each U i safely approximates W i : overapproximates uncontrollable behaviour underapproximates controllable behaviour ⇒ A will work for the real system. ...
BASYL can obtain models precise enough for controller synthesis.
Summary
Summary We presented a method to automatically compose an asynchronous system with black-box services. ...
doi:10.1007/978-3-642-31875-7_9
fatcat:qmqso476gzgntfzxkv5hpgua7m
Black-box Attacks on Deep Neural Networks via Gradient Estimation
2018
International Conference on Learning Representations
In this paper, we propose novel Gradient Estimation black-box attacks to generate adversarial examples with query access to the target model's class probabilities, which do not rely on transferability. ...
white-box attacks. ...
For an input with dimension d, the number of queries will be exactly 2d for a two-sided approximation. This may be too large when the input is high-dimensional. Random grouping. ...
dblp:conf/iclr/BhagojiH0S18
fatcat:dseqxxgyjfgt5mncgrbb7sw5ze
Explaining A Black-box By Using A Deep Variational Information Bottleneck Approach
2021
AAAI Conference on Artificial Intelligence
For each instance, VIBI selects key features that are maximally compressed about an input (briefness), and informative about a decision made by a black-box system on that input (comprehensive). ...
Briefness and comprehensiveness are necessary in order to provide a large amount of information concisely when explaining a black-box decision system. ...
VIBI then learns the explanations by only using input and output of the black-box system. ...
dblp:conf/aaai/BangXL0X21
fatcat:xn2zpss7bjbbrl7aa7rm4244qu
Safe Reinforcement Learning Using Black-Box Reachability Analysis
[article]
2022
arXiv
pre-print
Thus, we propose a Black-box Reachability-based Safety Layer (BRSL) with three main components: (1) data-driven reachability analysis for a black-box robot model, (2) a trajectory rollout planner that ...
Reinforcement learning (RL) is capable of sophisticated motion planning and control for robots in uncertain environments. ...
Algorithm 2: Black-box System Reachability [32] Input: initial reachable set R0 , actions (u j ) k+nplan j=k Parameter: state/action data D, noise zonotope W = Z(c w , G w ), Lipschitz constant L , and ...
arXiv:2204.07417v1
fatcat:vzrgtrclezd7lnnhl4xfvmph6a
Explaining a black-box using Deep Variational Information Bottleneck Approach
[article]
2019
arXiv
pre-print
For each instance, VIBI selects key features that are maximally compressed about an input (briefness), and informative about a decision made by a black-box system on that input (comprehensive). ...
Briefness and comprehensiveness are necessary in order to provide a large amount of information concisely when explaining a black-box decision system. ...
mimics the behaviour of the black-box system using the selected keys as the input. ...
arXiv:1902.06918v2
fatcat:fbg2dc6y4jfjvhuuhosruvsm2m
Quantification of Incertitude in Black Box Simulation Codes
[article]
2018
arXiv
pre-print
The methodology we apply is applicable to the problem of propagating input incertitudes through any simulation code treated as a "black box," i.e. a code for which the algorithmic details are either inaccessible ...
We have made the tools developed for this study freely available to the community. ...
Calder for previewing the manuscript. This research has made use of NASA's Astrophysics Data System Bibliographic Services. ...
arXiv:1805.09954v1
fatcat:63zgtotekbepvcvfackrc6xoxy
Hybrid MLP-RBF model structure for short-term internal temperature prediction in greenhouse environments
2013
2013 IEEE 14th International Symposium on Computational Intelligence and Informatics (CINTI)
The results show that the proposed method has better performance compared to the original physical-neural hybrid model if the input values are not far from the input range of the values used for training ...
This paper proposes a hybrid modeling method incorporating a multilayer perceptron neural network and a radial basis function neural network aimed to be more accurate on input regions not covered by training ...
ACKNOWLEDGMENT This research was partially supported by the ARTEMIS JU and the Hungarian National Development Agency (NFÜ) in frame of the R3-COP (Robust & Safe Mobile Co-operative Systems) project. ...
doi:10.1109/cinti.2013.6705225
fatcat:6cbotihlynfqlmxzgi4elhnur4
Emerging AI Security Threats for Autonomous Cars – Case Studies
[article]
2021
arXiv
pre-print
Adversaries can launch model extraction attacks for monetization purposes or step-ping-stone towards other attacks like model evasion. ...
With the emergence of AIoT technology, it is imperative to focus on the security of autonomous systems to make them robust and safe for their adoption. Fig Fig. 2. ...
We evaluate the effectiveness of both black-box and gray-box attack methods on the original model. In the black-box attack setting, queries are generated using random square blobs. ...
arXiv:2109.04865v1
fatcat:ntywbpeorjdo3exgm2bw3nyofe
Learning Performance-Oriented Control Barrier Functions Under Complex Safety Constraints and Limited Actuation
[article]
2024
arXiv
pre-print
approximation of the safe set. ...
Control Barrier Functions (CBFs) provide an elegant framework for designing safety filters for nonlinear control systems by constraining their trajectories to an invariant subset of a prespecified safe ...
The safe set X is denoted by the black box. ...
arXiv:2401.05629v1
fatcat:jz3maxjysbgargfhxwcldpwqby
The five Is: Key principles for interpretable and safe conversational AI
[article]
2021
arXiv
pre-print
AI that, unlike the currently popular black box approaches, is transparent and accountable. ...
At present, there is a growing concern with the use of black box statistical language models: While displaying impressive average performance, such systems are also prone to occasional spectacular failures ...
In cases where the input does not perfectly match any such intention sentence (which can happen, for example, with an approximation provided by a black box model), the agent should notify the user that ...
arXiv:2108.13766v1
fatcat:c2wxvq6okrdyvnyiv3messalxm
PAC Model Checking of Black-Box Continuous-Time Dynamical Systems
[article]
2020
arXiv
pre-print
for a given input. ...
In this paper we present a novel model checking approach to finite-time safety verification of black-box continuous-time dynamical systems within the framework of probably approximately correct (PAC) learning ...
Recently, scenario optimization was used to compute probably approximately safe inputs for a black-box system such that the system's final outputs fall within a safe range in [43] , and perform safety ...
arXiv:2007.10141v1
fatcat:3z3lrnilwfcfnbsmtvy3jkwfia
Exploiting Safety Constraints in Fuzzy Self-organising Maps for Safety Critical Applications
[chapter]
2004
Lecture Notes in Computer Science
The constrained FSOM has been termed a 'Safety Critical Artificial Neural Network' (SCANN) and preserves valuable performance characteristics for nonlinear function approximation problems. ...
Illustrations of potential benefits for real-world applications are also presented. ...
Limitations experienced from black-box analysis clearly highlight the need for improved neural models to allow compelling safety and performance arguments required for certification. ...
doi:10.1007/978-3-540-28651-6_39
fatcat:sylev6rdc5cenax5hqv2ufaiqy
« Previous
Showing results 1 — 15 out of 66,090 results