Cross-Dataset Generalization in Deep Learning

Deep learning has been extensively used in various fields, such as phase imaging, 3D imag ing reconstruction, phase unwrapping, and laser speckle reduction, particularly for complex problems that lack analytic models. Its data - driven nature allows for implicit construction of mathematical relationships within the network through training with abun dant data. However, a critical challenge in practical applications is the generalization issue, where a network trained on one dataset struggles to recognize an unknown target from a different dataset. In this study, we investigate imaging through scatteri ng media and discover that the mathematical relationship learned by the network is an approximation dependent on the training dataset, rather than the true mapping relationship of the model. W e demonstrate that enhancing the diversity of the training datas et can improve this approximation, thereby achieving generalization across different datasets, as the mapping relationship of a linear physical model is independent of inputs. This study elucidates the nature of generalization across different datasets and provides insights into the design of training datasets to ultimately address the generalization issue in various deep learning - based applications . Introduction The study of imaging through scattering media is a challenging and cutting - edge field. Scattering media are ubiquitous in everyday life, such as rough surfaces, clouds, fog, dust, water, and biological tissues. Image reconstruction through these media is p articularly important in areas such as transportation, military, and biomedicine .