Deep‐learning (DL) methods, in consideration of their excellence in dealing with highly complex structure–performance relationships for materials, are expected to become a new design paradigm for breakthroughs in material performance. However, in most cases, it is impractical to collect massive‐scale experimental data or open‐source theoretical databases to support training DL models with sufficient prediction accuracy. In a dataset consisting of 483 porous silicone rubber observations generated via ink‐writing additive manufacturing, this work demonstrates that constructing low‐dimensional, accurate descriptors is the prerequisite for obtaining high‐precision DL models based on small experimental datasets. On this basis, a unique convolutional bidirectional long short‐term memory model with spatiotemporal features extraction capability is designed, whose hierarchical learning mechanism further reduces the requirement for the amount of data by taking full advantage of data information. The proposed approach can be expected as a powerful tool for innovative material design on small experimental datasets, which can also be used to explore the evolutionary mechanisms of the structures and properties of materials under complex working conditions.

Filler dispersion is one of the key factors that significantly affect the performance of composite materials. Quantitative description of the dispersion state of fillers in composites and the internal relationship between the dispersion state and the macro‐mechanical properties of composites are very important for the performance design, prediction and processing process control of composites. In order to quantitatively and visually describe the dispersion state of inorganic fillers in matrix, research the relationship between fillers dispersion and composites properties. Here, by the fluorescent labeling technology of inorganic fillers and the surface organic modification technology, a modified fluorescently labeled calcium carbonate (MF‐CaCO3) with red fluorescence effect and good dispersibility was obtained, and polypropylene (PP)/MF‐CaCO3 composites with different loadings were prepared (the weight concentrations of MF‐CaCO3 in filler is 2, 5, 10, 15, and 20 wt%, respectively). The fractal dimension D value for evaluating the uniformity of filler dispersion has been successfully received, based on fractal analysis and fluorescent labeling‐laser scanning confocal microscope three‐dimensional (3D) visualization technology. Through the statistical results of experimental data, it was found that the value of fractal dimension D was obviously negatively correlated with the macroscopic mechanical properties of composite materials. When the dispersion of the filler was the best, that is, the fractal dimension D value took on a minimum and the composites had the best impact resistance. At the same time, this method based on fractal analysis and 3D visualization technology provides a new idea for quantitatively describing the dispersion state of fillers and its relationship with the mechanical properties of composites.