Methods and arrangements to identify activation profile context in training data

A method, the method comprising:

• receiving, by logic circuitry, a first activation profile associated with an event, the first activation profile to comprise an auxiliary output from a neural network based on activation of nodes in one or more layers of the neural network during inference to generate an inference output;
• correlating, by logic circuitry, the first activation profile against a second activation profile associated with a first training sample of training data, wherein the correlation comprises comparison of the first activation profile with multiple subsets of training samples in the training data;
• determining, by logic circuitry, that the first training sample is associated with the event based on the correlating;
• outputting, by logic circuitry, an indicator to identify the first training sample as being associated with the event;
• and
• generating, by graphics circuitry, a user interface to display a representation of the first training sample and a representation of a field sample for the event.

A computer program product comprising a non-transitory computer-readable medium, comprising instructions, which when executed by a processor cause the processor to perform operations, the operations to:

• access memory to determine a first activation profile associated with an event, the first activation profile based on activation of nodes in one or more layers of a neural network during inference to generate an inference output;
• correlate the first activation profile against a second activation profile associated with a first training sample of training data, wherein the correlation comprises comparison of the first activation profile with multiple subsets of training samples in the training data;
• determine that the first training sample is associated with the event based on correlation;
• output an indicator to identify the first training sample as being associated with the event;
• and
• generate a user interface to display a representation of the first training sample and a representation of a field sample for the event.

A system, the system comprising:

• a presentation device comprising processor circuitry and a display, the processor circuitry to select at least a first subset of training samples of training data to present to a user and the display to display a representation of the first subset of training samples and a representation of a field sample for an event;
• and
• logic circuitry coupled with the processor circuitry of the presentation device to:
• determine a first activation profile associated with the event, the first activation profile based on activation of nodes in one or more layers of a neural network during inference to generate an inference output;
• correlate the first activation profile against activation profiles associated with the first subset of training samples, wherein the correlation comprises comparison of the first activation profile with multiple subsets of training samples in the training data;
• determine that the first subset of training samples is associated with the event based on correlation;
• and
• output an indicator to the presentation device to identify the first subset of training samples as being associated with the event.