Double Descent and Intermittent Color Diffusion for Global Optimization and landscape exploration
arXiv:1706.052851 citationsh-index: 30
AI Analysis
For researchers in global optimization, this offers a new heuristic approach to escape local minima, though it is incremental in nature.
The paper introduces a global optimization method combining double descent and intermittent colored diffusion to explore smooth potential landscapes and find global minimizers. Numerical results demonstrate the method's performance.
In this work, we present a method to explore the landscape of a smooth potential in the search of global minimizers,combining a double-descent technique and a basin-escaping technique based on intermittent colored diffusion. Numerical results illustrate the performance of the method.