US 12,254,384 B2
Device and method to improve the robustness against ‘adversarial examples’
Jeremiah M. Cohen, Pittsburgh, PA (US); Frank Schmidt, Leonberg (DE); and Jeremy Zieg Kolter, Pittsburgh, PA (US)
Assigned to ROBERT BOSCH GMBH, Stuttgart (DE); and CARNEGIE MELLON UNIVERSITY, Pittsburgh, PA (US)
Appl. No. 17/272,435
Filed by Robert Bosch GmbH, Stuttgart (DE); and Carnegie Mellon University, Pittsburgh, PA (US)
PCT Filed Jan. 10, 2020, PCT No. PCT/IB2020/050176
§ 371(c)(1), (2) Date Mar. 1, 2021,
PCT Pub. No. WO2020/161546, PCT Pub. Date Aug. 13, 2020.
Claims priority of application No. 19155345 (EP), filed on Feb. 4, 2019.
Prior Publication US 2021/0319268 A1, Oct. 14, 2021
Int. Cl. G06F 18/2415 (2023.01); G06F 18/21 (2023.01); G06F 18/2431 (2023.01); G06N 7/01 (2023.01); G06N 20/00 (2019.01); G06V 10/774 (2022.01); G06V 10/776 (2022.01); G06V 10/82 (2022.01)
CPC G06N 20/00 (2019.01) [G06F 18/2193 (2023.01); G06F 18/2415 (2023.01); G06F 18/2431 (2023.01); G06N 7/01 (2023.01); G06V 10/774 (2022.01); G06V 10/776 (2022.01); G06V 10/82 (2022.01)] 10 Claims
OG exemplary drawing
 
1. A computer-implemented method for assessing a robustness of a smoothed classifier for classifying sensor signals received from a sensor, comprising the following steps:
providing an input signal depending on the sensor signal;
determining, by the smoothed classifier, a first value which characterizes a probability that the input signal, when subjected to noise, will be classified as belonging to a first class out of a predefined plurality of classes, wherein the first class is a most probable class;
determining, by the smoothed classifier, a second value which characterizes a probability that the input signal, when subjected to the noise, will be classified as belonging to a second class out of the predefined plurality of classes, wherein the second class is a second-most probable class; and
determining a robustness value on a first inverse value of a standard Gaussian cumulative distribution function at the first value and/or depending on a second inverse value of the standard Gaussian cumulative distribution function at the second value.