Abstract:
A robust approach to the construction of machine learning algorithms is considered, based on minimization
of robust finite sums of parametrized functions. It is based on the application of finite robust differentiable
aggregating summation functions that are stable against emissions.