Abstract | ||
---|---|---|
In this paper, we review the paradigm of inductive process modeling, which uses background knowledge about possible component processes to construct quantitative models of dynamical systems. We note that previous methods for this task tend to overfit the training data, which suggests ensemble learning as a likely response. However, such techniques combine models in ways that reduce comprehensibility, making their output much less accessible to domain scientists. As an alternative, we introduce a new approach that induces a set of process models from different samples of the training data and uses them to guide a final search through the space of model structures. Experiments with synthetic and natural data suggest this method reduces error and decreases the chance of including unnecessary processes in the model. We conclude by discussing related work and suggesting directions for additional research. |
Year | DOI | Venue |
---|---|---|
2005 | 10.1145/1102351.1102362 | ICML |
Keywords | Field | DocType |
quantitative model,process model,training data,model structure,inductive process modeling,natural data,process model induction,different sample,unnecessary process,additional research,ensemble learning,dynamic system | Training set,Computer science,Work in process,Process modeling,Dynamical systems theory,Artificial intelligence,Overfitting,Ensemble learning,Machine learning | Conference |
ISBN | Citations | PageRank |
1-59593-180-5 | 7 | 0.63 |
References | Authors | |
15 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Will Bridewell | 1 | 530 | 42.94 |
Narges Bani Asadi | 2 | 69 | 6.40 |
Pat Langley | 3 | 6471 | 1307.64 |
Ljupčo Todorovski | 4 | 276 | 15.25 |