The main and overarching objective of this research endeavor is to devise data-driven soft modeling tools that can provide fast, fairly accurate and computationally efficient predictions for: (1) the thermal conductivity of particulate composites (PC), and (2) the level of shielding provided by a porous structure from a high-yield explosion. The computational tools developed in this work constitute cheaper and faster alternatives to the relatively large and cumbersome numerical simulations or experimental testing campaigns that would otherwise be needed to make predictions of comparable ...