Emerging memristor-based computing has the potential to achieve higher efficiency over conventional von-Neumann architecture. Memristor-based neural networks employ bit-slicing to meet the high bit-precision demand. However, their accuracy is significantly degraded due to non-zero minimum conductance (Gmin) of memristors. This paper proposes an unbalanced bit-slicing scheme that provides higher sensing margin to more important bits for reducing the impact of non-zero Gmin. The unbalanced bit-slicing is assisted by 2's complement arithmetic to further improve the accuracy. Results show that the proposed scheme can achieve up to 89% and 44% accuracy improvement over the state-of-the-art for single-bit and two-bit configurations, respectively.