System and method for utilizing perturbation in a multimodal environment

A system including a machine learning network, comprising:

• an input interface configured to receive input data, wherein the input interface is connected to one or more sensors, wherein the one or more sensors includes a camera, radar, sonar, or microphone;
• and
• a processor, in communication with the input interface, wherein the processor is programmed to:
• receive the input data including information indicative of image information or sound information;
• select one or more batch samples from the input data;
• apply a perturbation object onto the one or more batch samples to create a perturbed sample;
• run the perturbed sample through the machine learning network in order to detect or classify the perturbed sample;
• and
• update the perturbation object in response to the machine learning network running the perturbed sample;
• and
• output the perturbation object in response to exceeding a convergence threshold, wherein the perturbation object is detectable by the machine learning network.

A computer-program product storing instructions which, when executed by a computer, cause the computer to:

• receive an input data;
• select one or more batch samples from the input data;
• apply a perturbation object onto the one or more batch samples to create a perturbed sample;
• run the perturbed sample through a machine learning network utilizing a function, wherein the function is configured to detect or classify the perturbed sample;
• train the machine learning network utilizing the function and in response to running the perturbed sample;
• and
• output a trained machine learning network.