<p>This paper is concerned with the bipartite synchronization for a class of coupled fuzzy memristive neural networks with cooperative-competitive relationships. To reflect the rigorous situations, proportional delays, as a kind of unbounded delays, is considered in both self circuits and broadcast process among coupled nodes. To ensure synchronous behavior under such a circumstance, a signed fuzzy impulsive control scheme integrating time-varying impulsive effects is proposed. For acquiring suitable impulsive intervals, a novel weighted memory-based self-triggered mechanism is elaborately designed. Notably, the self-triggered impulsive control protocol is upgraded to avoid postpone of control inputs caused by sparse monitoring mechanism by introducing a novel memory characteristic. Specially, by leveraging a weighted memory coefficient, the information close to current moment would be more dominant in decision-making of the triggering instants. With Filippov lemma, parameter variation approach and Lyapunov stability theorem, sufficient conditions for ensuring quasi-synchronization are derived. Finally, the simulation results prove the effectiveness and superiority of the given control method. Additionally, several comparative experiments are given to illustrate the necessity of weighted memory information.</p>