Blue Flamingo
0%

Multiple-Instance Learning

Posted on Valine:
What is MIL? What can we do with it?

What it is

  1. A form of weakly supervised learning
  2. Training instances are arranged in sets, called bags
  3. A label is provided for the entire bags, not for instances

Standard MIL assumption

  1. Negative bags do not contain positive
    instances.
  2. Positive bags may contain negative and
    positive instances.

Relaxed MIL assumption

  1. A bag is positive when it contains a sufficient number of positive
    instances.
  2. A bag is positive when it contains a certain combination of positive
    instances.
  3. Positive and negative bags differ by their instance distributions

Applications