| Did you know ... | Search Documentation: |
 | Pack ccprism -- prolog/ccprism/learn.pl |
- HOME
- DOWNLOAD
- DOCUMENTATION
- TUTORIALS
- Beginner▶
- Advanced▶
- Web applications▶
- Semantic web▶
- Graphics▶
- Machine learning▶
- External collections▶
- For packagers▶
- COMMUNITY
- COMMERCIAL
- WIKI
- converge(+C:convergence, +L:pred(-learner), -LL:list(float), +P1:sw_params, -P2:sw_params) is det
- Use L to create a predicate to do one step of learning, and then iterate
this until convergence, starting from P1 and ending with P2. History of
objective function values is returned in LL. Convergence C is of type:
convergence ---> abs(float); rel(float).
- learn(+Method:learn_method, +Stats:stats_method, +ITemp:number, +G:graph, -U:learner) is det
- learn(+Method:learn_method, +Stats:stats_method, +G:graph, -U:learner) is det
- Get update predicate for several EM-based parameter learning methods. learn/4 invokes
learn/5 with ITemp=1.0.
learn_method ---> ml; map(sw_params); vb(sw_params). stats_method ---> io(scaling); vit. scaling ---> lin; log. learner == pred(-float, +sw_params, -sw_params).