The past few years have seen the ability of cooperative Malware Detection Systems (MDS) to detect complex and unknown malware. In a cooperative setting, an MDS can consult other MDSs about suspicious malware and make a final decision using an aggregation mechanism. However, large delays may arise from both applying an aggregation mechanism and waiting to receive feedback from consulted MDSs. These shortcomings render the decisions produced by existing cooperative MDS approaches ineffective in real-time. To address the above-mentioned problem, we propose a deep learning-based cooperative MDS that efficiently exploits historical feedback data to foster proactive decision-making. More specifically, the proposed approach is based on Denoising Autoencoder (DA), which allows us to learn how to reconstruct complete MDSs' feedback from partial feedback. Our results show the effectiveness of the proposed framework on a real-life dataset.