Assessing Different Bayesian Neural Network Models for Militarized Interstate Dispute

Abstract

This article develops and compares two Bayesian neural network models, a more restrictive Bayesian framework using Gaussian approximation and a less restrictive one using a hybrid version of Markov Chain Monte Carlo method (HMC), for the prediction of militarized interstate disputes (MIDs). In addition, to compare and analyze different Bayesian models for international conflict, the authors introduce a new measurement to interpret the relative influence of the model variables on the MIDs. The results indicate that the Gaussian approximation and HMC models are not statistically different in their performance. However HMC correctly recognized a marginally higher number of militarized disputes whose classification is important for policy purpose. On the variable effect, both models indicate similar patter of influences, where the two key liberal variables, democracy and economic interdependence, produce a strong dynamic feedback loop among each other, which greatly increases or decreases the probability of MIDs.

Multinational Corporations and Political Violence: Trends and Forecasts

“…Random forests have been used and shown to be effective in several conflict forecasting applications (Hegre et al, 2019;Jones and Lupu, 2018;Colaresi and Mahmood, 2017;Hill and Jones, 2014). In addition to forecasting well, the algorithm can detect non-linear relationships in the data, and non-linearities are a well-known feature in conflict models (Muchlinski et al, 2016;Lagazio and Marwala, 2006;Beck et al, 2000). This should enable us to identify any relationship between terrorist acts against MNCs and MNC actions, should they exist.…”

Automatic Relevance Determination Bayesian Neural Networks for Credit Card Default Modelling

“…Due to the central limit theorem the Gaussian Approximation is expected to be a close approximation to the true posterior as the sample size becomes large [10].…”

“…The second drawback which applies to both ANNs and tree based models is that they give the probability of default but do not address the level of uncertainty behind such a probability -this does not aid in the reliability analysis and risk appetite assessment of the lending institution. In this work, we develop probabilistic formulations of ANNs through the Gaussian approximation [14,10] and Hybrid Monte Carlo [16] that will address the two shortcomings identified above.…”

“…The Bayesian formulation of ANNs using a Gaussian approximation to the posterior was first proposed by Mackay [11]. This method has shown superior performance in numerous applications of ANNs including in conflict analysis [10], energy consumption modelling [13] and wind power forecasting [14]. Hybrid Monte Carlo (HMC) was first applied as a technique for sampling from Bayesian Neural Network models (BNNs) in the seminal work of Neal [15].…”

Using Optimisation Techniques to Granulise Rough Set Partitions

“…Recent developments in the conflict literature has emphasised the importance of treating international conflicts as complex phenomena often displaying nonlinear and nonmonotonic patterns of interactions [71,72]. Various methods have been implemented and there are still efforts underway to study interstate interactions or militarised interstate disputes.…”