<p>It is critical to gain a quantifiable understanding of the inherent correlation between tensile and impact properties in order to effectively design alloys with high impact toughness based on available tensile property data. Machine learning methods excel at handling complex, nonlinear, and multidimensional problems, which is a feasible way to uncover the inherent correlation between tensile and impact properties. In the present study, the microstructure, mechanical properties (including tensile and impact properties), and fracture behavior of Sr- and La-modified Al-Si-Mg alloys in different aging states were investigated. Fractographic analyses were performed to study the different fracture mechanisms of quasi-statically deformed tensile specimens and dynamically deformed Charpy impact specimens. Based on the experimental data, an effective and interpretable machine learning model describing the inherent correlation between tensile and impact properties of Al-Si-Mg alloys was established. Further, Shapley additive explanations and sensitivity analyses were used to identify the significant features affecting the impact property. It can be concluded that the elongation of the Al-Si-Mg alloys plays a dominant role in influencing the impact properties. When it is not feasible to enhance both strength and ductility simultaneously, increasing ductility via a moderate reduction in strength is an effective strategy to improve impact toughness. This study offers a new perspective on heat treatment processes optimization and alloy design to develop alloys with superior impact toughness.</p>