Aerogels are known for their exceptional properties such as low mass density and high thermal insulation. These properties strongly depend on the fractal morphology and network structure of aerogels. To study this influence the diffusion limited cluster-cluster aggregation (DLCA) method and the Laguerre-Voronoi tessellation can be used for silica and nano porous open-cell aerogels, respectively [1,2]. In this contribution, we present a novel method based on deep symbolic regression (DSR) to improve the microstructural characterization in aerogel modeling. Symbolic regression is a data-driven method of machine learning that aims to identify an algebraic expression describing the relationship between input variables and a given output at best. Unlike neural networks, DSR provides a white box model that can be interpreted even for complex systems such as aerogels. In this work, insights are gained to understand better intricate relationships between input parameters and the resulting aerogels structures. The influence of some microstructural parameters as for example the particle radius or the pore size distribution is investigated. In this way, best material models for aerogels can be identified. REFERENCES [1] Pandit, P., Abdusalamov, R., Itskov, M., & Rege, A. (2024). Deep reinforcement learning for microstructural optimisation of silica aerogels. Scientific Reports, 14(1), 1511. [2] Chandrasekaran, R., Hillgärtner, M., Ganesan, K., Milow, B., Itskov, M., & Rege, A. (2021). Computational design of biopolymer aerogels and predictive modelling of their nanostructure and mechanical behaviour. Scientific Reports, 11(1), 10198.