Lattice structures often require extreme or spatially varying material properties for enhanced performance or complex functionalities. These architected materials, gaining prominence in aerospace, automotive, and medical industries, exhibit exceptional strength-to-weight ratios, thermal stability, and low density. Traditionally, achieving such tailored properties involves precise control of material composition and processing conditions, posing technical and cost challenges, especially when aiming for spatial variations. For instance, the hexagonal arrangement in honeycomb lattices combines strength and flexibility, making them ideal for applications like aircraft components, spacecraft parts, thermal protection systems, and energy storage. In modern structural applications, there is a growing need for structures to absorb energy efficiently while minimizing material use. Conventional approaches to designing periodic microstructural geometry may have reached saturation. Therefore, leveraging machine learning and artificial intelligence becomes crucial to advancing the mechanical properties of engineered lattice structures. This study introduces Gaussian Process Regression (GPR) based machine-learning models to predict failure band positions in honeycomb lattice structures. Utilizing the Sobol sampling technique, input parameters are obtained through random combinations of cell regularity, relative density (ranging from 3% to 9%), and elastoplastic material properties (within ± 20%). Concurrently, a sequential strengthening mechanism is proposed for identified failure bands in regular hexagonal lattices. As a result, there is a notable increase of up to 30% in energy absorption and ultimate stress for regular honeycomb lattices.