As double patterning techniques mature, they become the primary approaches enabling feature size scaling beyond 32nm. Although it is possible to print dense patterns by splitting the design into two masks, printability problems and pattern distortion remains a major concern. In this paper, we study the potential lithographic hotspots that may occur between the line ends in one-dimensional gridded designs obtained with Line-End Cut (LEC) method. We propose a post-placement hotspot detection and removal algorithm that perturbs the cell locations to eliminate all hotspots. Hotspot detection is performed using a pattern classifier based on machine learning techniques. Experimental results show that we can successfully eliminate all hotspots with excellent runtime efficiency and insignificant overhead on estimated wire lengths.