Learning and Verification of Feedback Control Systems using Feedforward Neural Networks

Dutta, Souradeep; Jha, Susmit; Sankaranarayanan, Sriram; Tiwari, Ashish

doi:10.1016/j.ifacol.2018.08.026

Cited by 67 publications

(53 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A neural network controller was trained for this system, using data-driven model predictive controller proposed in [11]. The trained network had 3 hidden layers, with 100 neurons in each layer making a total of 300 neurons.…”

Section: Sherlock-benchmark-9 (Tora)mentioning

confidence: 99%

ARCH-COMP20 Category Report: Artificial Intelligence and Neural Network Control Systems (AINNCS) for Continuous and Hybrid Systems Plants

Johnson¹,

Lopez²,

Musau³

et al.

EPiC Series in Computing

View full text Add to dashboard Cite

This report presents the results of a friendly competition for formal verification of continuous and hybrid systems with artificial intelligence (AI) components. Specifically, machine learning (ML) components in cyber-physical systems (CPS), such as feedforward neural networks used as feedback controllers in closed-loop systems are considered, which is a class of systems classically known as intelligent control systems, or in more modern and specific terms, neural network control systems (NNCS). We more broadly refer to this category as AI and NNCS (AINNCS). The friendly competition took place as part of the workshop Applied Verification for Continuous and Hybrid Systems (ARCH) in 2020. In the second edition of this AINNCS category at ARCH-COMP, four tools have been applied to solve seven different benchmark problems, (in alphabetical order): NNV, OVERT, ReachNN*, and VenMAS. This report is a snapshot of the current landscape of tools and the types of benchmarks for which these tools are suited. Due to the diversity of problems, lack of a shared hardware platform, and the early stage of the competition, we are not ranking tools in terms of performance, yet the presented results probably provide the most complete assessment of current tools for safety verification of NNCS.

show abstract

Section: Sherlock-benchmark-9 (Tora)mentioning

confidence: 99%

ARCH-COMP20 Category Report: Artificial Intelligence and Neural Network Control Systems (AINNCS) for Continuous and Hybrid Systems Plants

Johnson¹,

Lopez²,

Musau³

et al.

EPiC Series in Computing

View full text Add to dashboard Cite

show abstract

“…Sherlock uses an interleaving of gradient based local search technique, and mixed integer linear programming based global optimization solver to solve an optimization problem, for range analysis. Sherlock can be used to verify properties about closed loop systems, with deep neural networks as controllers as well [11]. This can be accomplished using ideas like template polyhedra and acceleration of reach set computation explained in [11].…”

Section: Participating Toolsmentioning

confidence: 99%

“…Sherlock can be used to verify properties about closed loop systems, with deep neural networks as controllers as well [11]. This can be accomplished using ideas like template polyhedra and acceleration of reach set computation explained in [11]. In, [10], the authors extended Sherlock to generate polynomial rules to characterize the local behavior of the network in a formally sound way.…”

Section: Participating Toolsmentioning

confidence: 99%

“…where w is a bounded error in the range [−0.01, 0.01]. A neural network controller was trained for this system, using data-driven model predictive controller proposed in [11]. The trained network had 3 hidden layers, with 100 neurons in each layer making a total of 300 neurons.…”

Section: Sherlock-benchmark-9 (Tora)mentioning

confidence: 99%

“…Sherlock. Reach sets of the ACC benchmark as obtained from Sherlock, along with 100 randomly chosen trajectories4.2.1 ACCThe reach sets for this example has been shown inFig 4.The initial sets shown in the figure correspond to x lead (0) ∈ [90, 91], x ego (0) ∈[10,11], v lead (0) ∈ [32,32.2], γ lead (0) = γ ego (0) = 0, v ego (0) ∈[30, 30.2]…”

mentioning

confidence: 99%

See 2 more Smart Citations

ARCH-COMP19 Category Report: Artificial Intelligence and Neural Network Control Systems (AINNCS) for Continuous and Hybrid Systems Plants

Lopez¹,

Musau²,

Tran³

et al.

EPiC Series in Computing

Self Cite

View full text Add to dashboard Cite

This report presents the results of a friendly competition for formal verification of continuous and hybrid systems with artificial intelligence (AI) components. Specifically, machine learning (ML) components in cyber-physical systems (CPS), such as feedforward neural networks used as feedback controllers in closed-loop systems are considered, which is a class of systems classically known as intelligent control systems, or in more modern and specific terms, neural network control systems (NNCS). For future iterations, we more broadly refer to this category as AI and NNCS (AINNCS). The friendly competition took place as part of the workshop Applied Verification for Continuous and Hybrid Systems (ARCH) in 2019. In the first edition of this AINNCS category at ARCH-COMP, three tools have been applied to solve five different benchmark problems, (in alphabetical order): NNV, Sherlock, and Verisig. This report is a snapshot of the current landscape of tools and the types of benchmarks for which these tools are suited. Due to the diversity of problems and that this is the first iteration of this category, we are not ranking tools in terms of performance, yet the presented results probably provide the most complete assessment of tools for the safety verification of NNCS.

show abstract

Trinity: Trust, Resilience and Interpretability of Machine Learning Models

Jha

Jalaian

Roy

et al. 2021

Game Theory and Machine Learning for Cyber Security

View full text Add to dashboard Cite

Learning and Verification of Feedback Control Systems using Feedforward Neural Networks

Cited by 67 publications

References 9 publications

ARCH-COMP20 Category Report: Artificial Intelligence and Neural Network Control Systems (AINNCS) for Continuous and Hybrid Systems Plants

ARCH-COMP20 Category Report: Artificial Intelligence and Neural Network Control Systems (AINNCS) for Continuous and Hybrid Systems Plants

ARCH-COMP19 Category Report: Artificial Intelligence and Neural Network Control Systems (AINNCS) for Continuous and Hybrid Systems Plants

Trinity: Trust, Resilience and Interpretability of Machine Learning Models

Contact Info

Product

Resources

About