In this paper, we consider the binary quadratic programming problems (\(BQP\)). The unconstrained \(BQP\) is known to be NP-hard and has many practical applications like signal processing, economy, management and engineering. Due to this reason, many algorithms have been proposed to improve its effectiveness and efficiency. In this paper, we propose a novel algorithm based on the basic algorithm proposed in [1], [2, 3] to solve problem \(BQP\) with \(Q\) being a seven-diagonal matrix. It is shown that the proposed algorithm has good performance and high efficiency. To further improve its efficiency, the neural network implementation is realized.