System and method for estimating uncertainty of the decisions made by a supervised machine learner
AssigneeHRL Laboratories, LLC
InventorsSoheil Kolouri, Heiko Hoffmann
Life of this patent
Estimating the term…
Abstract
Described is a system for controlling autonomous platform. Based on an input image, the system generates a motor control command decision for the autonomous platform. A probability of the input image belonging to a set of training images is determined, and a reliability measure for the motor control command decision is generated using the determined probability. An exploratory action is performed when the reliability measure is above a predetermined threshold. Otherwise, an exploitation action corresponding with the motor control command decision is performed when the reliability measure is below a predetermined threshold.
Claims21
A system for controlling an autonomous platform, the system comprising:
- one or more processors and a non-transitory computer-readable medium having executable instructions encoded thereon such that when executed, the one or more processors perform operations of:
- generating a motor control command decision for the autonomous platform based on an input image;
- determining a probability of the input image belonging to a set of training images;
- training a decision module with the set of training images via supervised learning or reinforcement learning, wherein the decision module generates the motor control command decision;
- training an uncertainty module with the set of training images in an unsupervised manner, wherein the uncertainty module determines the probability of the input image belonging to the set of training images,
- wherein the uncertainty module is trained in parallel with the decision module;
- generating a reliability measure for the motor control command decision using the determined probability;
- and
- performing an exploratory action when the reliability measure is above a predetermined threshold;
- otherwise
- performing an exploitation action corresponding with the motor control command decision when the reliability measure is below a predetermined threshold.
A computer implemented method for controlling an autonomous platform, the method comprising an act of:
- causing one or more processers to execute instructions encoded on a non-transitory computer-readable medium, such that upon execution, the one or more processors perform operations of:
- generating a motor control command decision for the autonomous platform based on an input image;
- determining a probability of the input image belonging to a set of training images;
- training a decision module with the set of training images via supervised learning or reinforcement learning, wherein the decision module generates the motor control command decision;
- training an uncertainty module with the set of training images in an unsupervised manner, wherein the uncertainty module determines the probability of the input image belonging to the set of training images,
- wherein the uncertainty module is trained in parallel with the decision module;
- generating a reliability measure for the motor control command decision using the determined probability;
- and
- performing an exploratory action when the reliability measure is above a predetermined threshold;
- otherwise
- performing an exploitation action corresponding with the motor control command decision when the reliability measure is below a predetermined threshold.
A computer program product for controlling an autonomous platform, the computer program product comprising:
- computer-readable instructions stored on a non-transitory computer-readable medium that are executable by a computer having one or more processors for causing the processor to perform operations of:
- generating a motor control command decision for the autonomous platform based on an input image;
- determining a probability of the input image belonging to a set of training images;
- training a decision module with the set of training images via supervised learning or reinforcement learning, wherein the decision module generates the motor control command decision;
- training an uncertainty module with the set of training images in an unsupervised manner, wherein the uncertainty module determines the probability of the input image belonging to the set of training images,
- wherein the uncertainty module is trained in parallel with the decision module;
- generating a reliability measure for the motor control command decision using the determined probability;
- and
- performing an exploratory action when the reliability measure is above a predetermined threshold;
- otherwise
- performing an exploitation action corresponding with the motor control command decision when the reliability measure is below a predetermined threshold.
Dependencies are inferred from the claim text.