In the single-particle approach, a large number of randomly oriented projection images of a molecule are assembled into a consistent 3D view. To reach a high resolution, hundreds of thousands of such particle views must be selected from electron micrographs.

To improve the performance of such particle picking methods, we employ Support Vector Machines. The SVMs are trained using an example data set of correct and wrong picks and "learn" the characteristics of the picking process from these examples. First result indicate that SVMs can significantly increase picking performance (as measured by the number of false positives vs. true positives).