First-order methods almost always avoid strict saddle points
Mathematical Programming, pp. 1-27, 2019.
We establish that first-order methods avoid strict saddle points for almost all initializations. Our results apply to a wide variety of first-order methods, including (manifold) gradient descent, block coordinate descent, mirror descent and variants thereof. The connecting thread is that such algorithms can be studied from a dynamical sys...More
Get fulltext within 24h
Full Text (Upload PDF)
PPT (Upload PPT)