Abstract. In this paper, we pose a novel research problem for machine learning that involves constructing a process model from continuous data. We claim that casting learned knowledge in terms of processes with associated equations is desirable for scientific and engineering domains, where such notations are commonly used. We also argue that existing induction methods are not well suited to this task, although some techniques hold partial solutions. In response, we describe an approach to learning process models from time-series data and illustrate its behavior in three domains. In closing, we describe open issues in process model induction and encourage other researchers to tackle this important problem.

Abstract In this paper, we pose a novel research problem for machine learning that involves constructing a process model from continuous data. We claim that casting learned knowledge in terms of processes with associated equations is desirable for scientific and engineering domains, where such notations are commonly used. We also argue that existing induction methods are not well suited to this task, although some techniques hold partial solutions. In response, we describe an approach to learning process models from time-series data and illustrate its behavior in three domains. In closing, we describe open issues in process model induction and encourage other researchers to tackle this important problem.

Ecological experts use qualitative models to capture and communicate their domain knowledge, but creating qualitative models is a complex task that requires expertise ecologists do not normally possess. This thesis presents an improved version of the automatic model building algorithm that supports model building efforts by automatically creating qualitative process models based on a description of system behavior. The improved algorithm is applicable to systems containing interacting processes. The algorithm is tested against models created by experts, showing the learned models to be conceptually correct.