Loading...
Please wait, while we are loading the content...
Similar Documents
A Scalable Method for Classiier Knowledge Reuse
| Content Provider | Semantic Scholar |
|---|---|
| Author | Bollacker, Kurt D. Ghosh, Joydeep |
| Copyright Year | 1997 |
| Abstract | Just as a person's lifelong experience helps him/her in new classiication tasks, it would be useful to leverage the knowledge in previously trained artiicial classiiers in learning future classiication tasks. This knowledge may improve classiier performance, speed up learning, and assist in problem decomposition. We present a maximum likelihood method for classiier knowledge reuse that is novel in its scalability with the quantity of classiiers reused and in its ability to incorporate diierent classiier architectures. Also, we describe a mutual information based relevance criterion for previously trained classiiers. Results from application of this method and criterion to public domain data sets demonstrate their usefulness. |
| File Format | PDF HTM / HTML |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
