Particle-laden composites are typical thermal interfacial materials (TIMs) in the electronic applications, which are widely used in the electron packaging fields. The effective thermal conductivity (effective TC) of the particle-laden composites is dominant by the particle-matrix and particle-particle interfacial thermal resistance (ITR). The reliable identification of ITR is essential for the accurate prediction of TC of the composites, which has potential in the design of TIMs. In this work, we propose an efficient strategy to identify the interfacial thermal resistance in the particle-laden composites combining the numerical simulation, high-throughput computation, machine learning algorithm and simple experimental measurement. Firstly, the high-throughput computation is conducted based on the numerical modeling of the standard samples, in which the input parameters are ITRs in the composites. Afterwards, a prototypical function-based machine learning strategy is employed on the database to describe the numerical relation between the effective TC and the input parameters. Finally, comparing the numerical predictions from the machine learning model with the experimental measurement of the effective TC, a high-throughput screening of the ITRs is executed for the identification of their values. The reliability of the strategy is validated by an example of Al2O3-AlN/silicone composites, showing that the particle-particle ITR is higher than particle-matrix ITR.