Sign inSign in
ConceptAI & ML

Data poisoning

A failure mode where evaluation datasets or training methods are artificially constructed to be IID in a way that does not represent real-world deployment.
Watch it explained

Know first

Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.

IID assumption1Loss function1Model evaluation1Pytorch labels1

Opens up

Concepts that name this one as a prerequisite.

Lecturers

2

Each in their own terms, from the easiest level.

  • Ashwinee Panda

    Caltech · Research

    Inserting a crafted canary into a target data point and isolating the loss signal to force the model to memorize the canary without affecting overall performance.
    • 14:10Worst-Case Membership Inference of Language Models

Lectures that assume it

1

Connections

5

The lectures that define it or assume it, and the concepts next to it.

2 lecturers. Definitions are written by AI from each lecture’s transcript. A time with a play mark starts the lecture at the moment the lecturer says it.