Out-of-distribution detection with projection of gradients

A system for detecting out-of-distribution data for a neural network, the system comprising:

• a processor;
• and
• memory having instructions that, when executed by the processor, cause the processor to perform the following:
• receive a training dataset, wherein the training dataset includes in-distribution data including image data associated with one or more images;
• train a neural network on the in-distribution data, wherein the neural network has a plurality of layers;
• generate a subspace of in-distribution data of the training dataset based on representations obtained from a selected layer of the neural network trained with the in-distribution data;
• receive image data associated with a sample image for the neural network;
• execute the neural network with the image data associated with the sample image to determine, for the selected layer, a gradient associated with the sample image;
• project the gradient for the selected layer into the subspace to derive a projection of the gradient;
• and
• determine that the image data associated with the sample image is out of distribution (OOD) based on a magnitude of the projection of the gradient.

A non-transitory computer-readable medium having stored thereon computer-readable instructions that, when executed by a processor, cause the processor to execute a method for detecting out-of-distribution data for a neural network, the method comprising:

• receiving a training dataset, wherein the training dataset includes in-distribution data including image data associated with one or more images;
• training a neural network on the in-distribution data, wherein the neural network has a plurality of layers;
• generating a subspace of in-distribution data of the training dataset based on representations obtained from a selected layer of the neural network trained with the in-distribution data;
• receiving image data associated with a sample image for the neural network;
• executing the neural network with the image data associated with the sample image to determine, for the selected layer, a gradient associated with the sample image;
• projecting the gradient for the selected layer into the subspace to derive a projection of the gradient;
• and
• determining that the image data associated with the sample image is out of distribution (OOD) based on a magnitude of the projection of the gradient.