Loading...
Please wait, while we are loading the content...
Similar Documents
Feature Extraction Labeled Training Data Classifier Training Classifier Model Unknown Data Classification Results Feature Extraction
| Content Provider | Semantic Scholar |
|---|---|
| Abstract | Application domains include a toolbox of different techniques and algorithms perfected over time. Much of this knowledge, however, still remains locked up in the domain literature and not readily accessible to people not familiar with the application area. We believe patterns can help us expose the domain knowledge to others not familiar with the depth of the domain and aid the creation of application frameworks that will let software writers (not necessarily domain experts) create interesting applications. These application patterns also serve to show how application domains feed the structural and computational patterns in OPL. In addition to pattern mining from individual examples, application patterns will help codify common practices and help programming framework development tailored to particular application areas. The Extract Classify pattern is used in a variety of contexts and domains. In general, the problem centers around understanding data. An overview of the problem is presented in figure 1. In the Extract Classify pattern, we start with Figure 1: The Extract Classify Pattern some labeled training data, which has been annotated with results known to be correct. We extract features from this data, and then train a classifier based on these features and our known labels. This training process produces a model. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | http://parlab.eecs.berkeley.edu/wiki/_media/patterns/extractclassify.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
