To improve the linear contour accuracy of 2X/Y gantry linear feed axes, the research of the literature focuses on the straightness accuracy measurement, evaluation and error compensation of the axes. Firstly, The cause of linear contour error in interpolation process and its compensation complexity were analyzed for a 2X/Y gantry feed axis driven by linear motors. Then a learning algorithm based compensation method was applied to increase the contour accuracy for this type of gantry axis.In proposed method, the 2D time function of a laser interferometer was utilized to acquire real-time interpolation error data and the least square method was taken to evaluate and determine the ideal linear equation. A model based on least square support vector regression technique was established to recognize the error. With the support vectors after learning process, the real-time compensation values were acquired through the model regression calculation. Finally, the compensation output strategy and corresponding realization system were also proposed. To demonstrate the procedure of the proposed approach, an experiment was conducted on the self-construction 2X/Y axis feeding platform. The result shows that the combination technique can compensate the interpolation error and can increase the straightness accuracy by 53%.