14 March 2024 in IEEE Transactions on Neural Networks and Learning Systems
The great success of deep learning poses an urgent challenge to establish the theoretical basis for its working mechanism. Recently, research on the convergence of deep neural networks (DNNs) has made great progress. However, the existing studies are based on the assumption that the samples are independent, which is too strong to be applied to many real-world scenarios. In this brief, we establish a fast learning rate for the empirical risk minimization (ERM) on DNN regression with dependent samples, and the dependence is expressed in terms of geometrically strongly mixing sequence. To the best of our knowledge, this is the first convergence result of DNN methods based on mixing sequences. This result is a natural generalization of the independent sample case.
30 March 2018 in Mathematical Methods in the Applied Sciences
1 October 2020 in Journal of Differential Equations
1 January 2018 in Colloquium Mathematicum
We prove a unique continuation theorem for functions $u$ vanishing to infinite order in the space-time variables at $(0,0)$ and satisfying the inequality $|\varDelta u+\partial _t u|\le V(x,t)|u|+W(x,t)|\nabla u|$ for some unbounded time-dependent $V$ and