Active Learning of Continuous-time Bayesian Networks through Interventions
This work addresses the bottleneck of high experimental costs in fields like natural and social sciences by enabling more efficient learning of CTBNs through interventions.
The authors tackled the problem of learning Continuous-time Bayesian Networks (CTBNs) with limited experimental data by proposing a variational approximation for experimental design, which reduces computational costs in high-dimensional settings and demonstrates performance on synthetic and real-world data.
We consider the problem of learning structures and parameters of Continuous-time Bayesian Networks (CTBNs) from time-course data under minimal experimental resources. In practice, the cost of generating experimental data poses a bottleneck, especially in the natural and social sciences. A popular approach to overcome this is Bayesian optimal experimental design (BOED). However, BOED becomes infeasible in high-dimensional settings, as it involves integration over all possible experimental outcomes. We propose a novel criterion for experimental design based on a variational approximation of the expected information gain. We show that for CTBNs, a semi-analytical expression for this criterion can be calculated for structure and parameter learning. By doing so, we can replace sampling over experimental outcomes by solving the CTBNs master-equation, for which scalable approximations exist. This alleviates the computational burden of sampling possible experimental outcomes in high-dimensions. We employ this framework in order to recommend interventional sequences. In this context, we extend the CTBN model to conditional CTBNs in order to incorporate interventions. We demonstrate the performance of our criterion on synthetic and real-world data.