Multiple-instance learning with structured bag models
<p>Traditional approaches to Multiple-Instance Learning (MIL) operate under the assumption that the instances of a bag are generated independently, and therefore typically learn an instance-level classifier which does not take into account possible dependencies between instances. This assumpti...
| Main Authors: | , |
|---|---|
| Format: | Conference item |
| Language: | English |
| Published: |
Springer
2011
|
| _version_ | 1826315009412562944 |
|---|---|
| author | Warrell, J Torr, PHS |
| author_facet | Warrell, J Torr, PHS |
| author_sort | Warrell, J |
| collection | OXFORD |
| description | <p>Traditional approaches to Multiple-Instance Learning (MIL) operate under the assumption that the instances of a bag are generated independently, and therefore typically learn an instance-level classifier which does not take into account possible dependencies between instances. This assumption is particularly inappropriate in visual data, where spatial dependencies are the norm. We introduce here techniques for incorporating MIL constraints into Conditional Random Field models, thus providing a set of tools for constructing <em>structured bag</em> models, in which spatial (or other) dependencies are represented. Further, we show how Deterministic Annealing, which has proved a successful method for training non-structured MIL models, can also form the basis of training models with structured bags. Results are given on various segmentation tasks.</p> |
| first_indexed | 2024-12-09T03:16:06Z |
| format | Conference item |
| id | oxford-uuid:4ccc13ca-adff-46be-9f7e-3498481a1ce2 |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-12-09T03:16:06Z |
| publishDate | 2011 |
| publisher | Springer |
| record_format | dspace |
| spelling | oxford-uuid:4ccc13ca-adff-46be-9f7e-3498481a1ce22024-10-24T14:15:57ZMultiple-instance learning with structured bag models Conference itemhttp://purl.org/coar/resource_type/c_5794uuid:4ccc13ca-adff-46be-9f7e-3498481a1ce2EnglishSymplectic ElementsSpringer2011Warrell, JTorr, PHS<p>Traditional approaches to Multiple-Instance Learning (MIL) operate under the assumption that the instances of a bag are generated independently, and therefore typically learn an instance-level classifier which does not take into account possible dependencies between instances. This assumption is particularly inappropriate in visual data, where spatial dependencies are the norm. We introduce here techniques for incorporating MIL constraints into Conditional Random Field models, thus providing a set of tools for constructing <em>structured bag</em> models, in which spatial (or other) dependencies are represented. Further, we show how Deterministic Annealing, which has proved a successful method for training non-structured MIL models, can also form the basis of training models with structured bags. Results are given on various segmentation tasks.</p> |
| spellingShingle | Warrell, J Torr, PHS Multiple-instance learning with structured bag models |
| title | Multiple-instance learning with structured bag models
|
| title_full | Multiple-instance learning with structured bag models
|
| title_fullStr | Multiple-instance learning with structured bag models
|
| title_full_unstemmed | Multiple-instance learning with structured bag models
|
| title_short | Multiple-instance learning with structured bag models
|
| title_sort | multiple instance learning with structured bag models |
| work_keys_str_mv | AT warrellj multipleinstancelearningwithstructuredbagmodels AT torrphs multipleinstancelearningwithstructuredbagmodels |