Skrzypczak, Glenn and Uhrmacher, Adelinde M. (2026) Towards Benchmarking Methods for Learning Chemical Reaction Networks from Time Series Data. In: 40th ACM SIGSIM Conference on Principles of Advanced Discrete Simulation, 24-26 June 2026, Vienna, Austria. Proceedings, published by Association for Computing Machinery (ACM), New York, NY, USA, pp. 195-196.
Full text not available from this repository.Abstract
Automatically learning chemical reaction networks (CRNs) from time series concentration data yields models with a causal structure that can be interpreted and refined, e.g., by further experiments, providing valuable insights into the underlying mechanisms [7]. CRNs represent processes as sets of reactions, where reactants are transformed into products at specific rates given by rate laws. This makes them particularly useful for modeling dynamic systems in fields such as biology and chemistry. A recent survey by Kreikemeyer and Uhrmacher [5] identified 68 distinct CRN-learning approaches. The survey shows that most publications are evaluated only on a few toy examples. This scarcity of diverse test cases, combined with divergent datasets, assumptions, and metrics, coupled with infrequent comparisons to prior work, makes it difficult to verify reported gains and obscure overall progress.
| Item Type: | Conference or Workshop Item (Paper) |
|---|
