We consider the basic problem of refi tting a time series over a finite period of time and formulate it as a stochastic dynamic program. By changing the underlying Markov decision process we are able to obtain a model that at optimality considers historical data as well as forecasts of future...

The Household Finance and Consumption Survey (HFCS) provides valuable information for the monetary policy and financial stability purposes. The dataset shows, however, inconsistencies with National Account (NtlA) statistics, as the aggregated HFCS micro data do usually not match the...

We propose a new method for solving high-dimensional dynamic programming problems and recursive competitive equilibria with a large (but finite) number of heterogeneous agents using deep learning. The "curse of dimensionality" is avoided due to four complementary techniques: (1) exploiting...

We propose an optimal architecture for deep neural networks of given size. The optimal architecture obtains from maximizing the minimum number of linear regions approximated by a deep neural network with a ReLu activation function. The accuracy of the approximation function relies on the neural...

Calibration of financial models can have more than one local minima present, requiring the use of global optimization techniques to properly calibrate them. In general, calibrating with a global optimizer will be a slow operation. An artificial neural network, properly trained, can replicate the...

This paper proposes a novel method of global optimization based on host-parasite co-evolution. It also develops a Fortran-77 code for the algorithm. The algorithm has been tested on 100 benchmark functions (of which the results of 32 relatively harder problems have been reported). In its search...

This note is intended to demonstrate that median (a well-known measure of central tendency) is a weighted arithmetic mean of all sample observations and such (non-trivial) weights may be computed by an optimization method such as the host-parasite co-evolutionary algorithm

This paper addresses the problem of utility maximization under uncertain parameters. In contrast with the classical approach, where the parameters of the model evolve freely within a given range, we constrain them via a penalty function. We show that this robust optimization process can be...

In this paper we examine feed-forward neural networks using genetic algorithms in the training process instead of error backpropagation algorithm. Additionally real encoding is preferred to binary encoding as it is more appropriate to find the optimum weights. We use learning and momentum rates...