Principal Geodesic Analysis (PGA) overcomes the linearity constraints of Principal Component Analysis (PCA), making it better suited for handling non-linear data residing on a low-dimensional manifold.