Skip to yearly menu bar Skip to main content


Toward Deeper Understanding of Neural Networks: The Power of Initialization and a Dual View on Expressivity

Amit Daniely · Roy Frostig · Yoram Singer

Area 5+6+7+8 #174

Keywords: [ Regularization and Large Margin Methods ] [ Learning Theory ] [ Kernel Methods ]


We develop a general duality between neural networks and compositional kernel Hilbert spaces. We introduce the notion of a computation skeleton, an acyclic graph that succinctly describes both a family of neural networks and a kernel space. Random neural networks are generated from a skeleton through node replication followed by sampling from a normal distribution to assign weights. The kernel space consists of functions that arise by compositions, averaging, and non-linear transformations governed by the skeleton's graph topology and activation functions. We prove that random networks induce representations which approximate the kernel space. In particular, it follows that random weight initialization often yields a favorable starting point for optimization despite the worst-case intractability of training neural networks.

Live content is unavailable. Log in and register to view live content