Research on Algorithm Optimization Driven by Big Data

To address the performance degradation and limited adaptability of traditional optimization algorithms in big data environments, this paper proposes an optimization framework that integrates data representation, feedback adjustment, and model fusion. Through feature compression and structural modeling, data characteristics are embedded in the objective function expression, a search space compression mechanism with dynamic feedback capability is constructed, and the robustness of the algorithm is improved by combining parameter adaptation strategies. In the scenario of multi-source heterogeneous data, an integrated optimization scheme is further introduced to improve the generalization ability. Comparative and ablation experiments are carried out based on three types of real data sets, and the superior performance of the proposed method in terms of accuracy, convergence and resource control is systematically verified.