<p>In this paper, we investigate the neural-network-based cooperative resilient nonconvex optimization control issue for high-order nonlinear multi-agent systems (MASs) in the presence of denial-of-service (DoS) attacks. Compared to existing distributed optimization algorithms, a novel control framework for cooperative resilient nonconvex optimization is adopted, which is structured by designing a resilient distributed nonconvex optimization algorithm, developing a smooth-like trajectory, and proposing a neural network backstepping-based controller. Specifically, a resilient distributed nonconvex optimization algorithm is first proposed to ensure convergence under DoS attacks, where the global cost function satisfies the Polyak-Łojasiewicz (P-Ł) condition, which does not require the global cost function to be convex, and the global minimizer is not necessarily unique. Then, a smooth-like trajectory is constructed via Hermite interpolation to generate a new variable with well-defined high-order time derivatives. Furthermore, a neural network backstepping-based scheme is designed for high-order nonlinear MASs, ensuring that the system output tracks the optimal value. Finally, a simulation example is presented to verify the effectiveness of our proposed method.</p>