2022-10-17
New perspective on the convergence to a global solution of finite-sum optimization
Publication
Publication
Deep neural networks have shown great success in many machine learning tasks. Their training is challenging since the loss surface of the network architecture is generally non-convex, or even non-smooth. We propose a reformulation of the minimization problem allowing for a new recursive algorithmic framework. By using bounded style assumptions, we prove convergence to an \epsilon-(global) minimum using O(1/\epsilon^3) gradient computations. Our theoretical foundation motivates further study, implementation, and optimization of the new algorithmic framework and further investigation of its non-standard bounded style assumptions.
Additional Metadata | |
---|---|
IBM Research, Thomas J. Watson Research Center, USA | |
Organisation | Computer Security |
Nguyen, L., Tran, T., & van Dijk, M. (2022). New perspective on the convergence to a global solution of finite-sum optimization. In Informs Annual Meeting. |
Additional Files | |
---|---|
View at homepage |