Multiple-instance learning: Difference between revisions

Revision as of 00:10, 1 March 2008

Revision as of 00:28, 1 March 2008

Line 1:

{{orphan|date=December 2007}}

'''Multiple-instance learning''' is a variation on [[supervised learning]], where the task is to learn a concept from data consisting of a sequence of instances, each labeled as positive or negative, and each described as a [[set]] of [[Coordinate vector|vectors]]. The instance is positive if at least one of the vectors in its set lies within the intended concept, and negative if none of the vectors lies within the concept; the task is to learn an accurate description of the concept from this information.

{{deadend|date=December 2007}}

Multiple-instance learning was originally proposed under this name by {{harvtxt|Dietterich|Lathrop|Lozano-Pérez|1997}}, but earlier examples of similar research exist, for instance in the work on [[handwriting|handwritten]] [[digit]] [[optical character recognition|recognition]] by {{harvtxt|Keeler|Rumelhart|Leow|1990}}.

Numerous researchers have worked on adapting classical classification techniques, such as [[support vector machines]] or [[boosting]], to work within the context of multiple-instance learning.

Multiple-instance learning is a variation on supervised learning, where the task is to learn a concept given positive and negative bags of instances. Each bag may contain many instances, but a bag is labeled positive even if only one of the instances in it falls within the concept. A bag is labeled negative only if all the instances in it are negative.

==References==

The idea for multiple instance learning was originally proposed 1990 for handwritten digit recognition by Keeler, et.al. Keeler's approach was called Integrated Segmentation and Recognition (ISR).

*{{citation

| first1 = Thomas G. | last1 = Dietterich

| first2 = Richard H. | last2 = Lathrop

| first3 = Tomás | last3 = Lozano-Pérez

| title = Solving the multiple instance problem with axis-parallel rectangles

| journal = Artificial Intelligence

| volume = 89 | issue = 1–2 | year = 1997 | pages = 31–71 | doi = 10.1016/S0004-3702(96)00034-3}}.

*{{citation

Another relevant example of MIL is the Diverse Density approach of Maron. Diverse Density uses the Noisy OR generative model to explain the bag labels. A gradient-descent algorithm is used to find the best point in input space that explains the positive bags.

| first1 = James D. | last1 = Keeler

| first2 = David E. | last2 = Rumelhart

| first3 = Wee-Kheng | last3 = Leow

| title = Integrated segmentation and recognition of hand-printed numerals

| Proceedings of the 1990 Conference on Advances in Neural Information Processing Systems (NIPS 3)

| year = 1990 | pages = 557–563}}.

[[Category:Machine learning]]

A lot of researches is being conducted to adapt classical classifiers, such as SVM or Boost, to work with MIL.