Our Services

Discover rare events for purposes of optimization, testing and verification, and generation of synthetic ML-training data.

Discover Rare Events Pertaining to Your Domain

Examples of rare events:

Paths taken by autonomous vehicles that result in a collision. The discovery of such paths is the backbone of verification.

Paths taken by a plurality of agents that satisfy a collection of individual and group constraints.

Resource allocation that satisfies a knapsack set of constraints.

Path finding in the presence of uncertainties.

Create Synthetic ML-Training Data Based on Rare-Events

Employ a patent-pending technique called HybridPair to discover a high variance train/test data set consisting of 0/1 labeled data items of much higher quality than those discoverable using Monte Carlo methods.

Related Publications

  • M. Litton, D. Drusinsky, L. Bridget and J. B. Michael, "Machine-Learned Correctness Properties, Runtime Verification, and Advance-Warning Oracles for Autonomous Systems," in Computer, vol. 57, no. 10, pp. 118-130, Oct. 2024, doi: 10.1109/MC.2024.3433669. keywords: {Runtime;Uncertainty;Autonomous systems;Measurement uncertainty;Machine learning;Predictive models},

  • D. Drusinsky, M. Litton and J. B. Michael, "Machine-Learned Verification and Advance Notice Oracles for Autonomous Systems," in Computer, vol. 56, no. 7, pp. 121-130, July 2023, doi: 10.1109/MC.2023.3265732. keywords: {Autonomous systems;Cyber-physical systems},

  • Machine-Learned Correctness Properties, Runtime Verification, and Advance-Warning Oracles for Autonomous Systems

    Oct.2024,pp. 118-130,vol. 57

    DOI Bookmark: 10.1109/MC.2024.3433669

  • Multiagent Pathfinding Under Rigid, Optimization, and Uncertainty Constraints

    Jul.2021,pp. 111-118,vol. 54

    DOI Bookmark: 10.1109/MC.2021.3074264