Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Neural Network-Based Parameter Estimation of a Labour Market Agent-Based Model

M Lopes Alves, Joel Dyer, Doyne Farmer, Michael Wooldridge, Anisoara Calinescu

TL;DR

A state-of-the-art simulation-based inference framework that uses neural networks (NN) for parameter estimation is evaluated and it is demonstrated that the NN-based approach recovers the original parameters when evaluating posterior distributions across various dataset scales and improves efficiency compared to traditional Bayesian methods.

Abstract

Agent-based modelling (ABM) is a widespread approach to simulate complex systems. Advancements in computational processing and storage have facilitated the adoption of ABMs across many fields; however, ABMs face challenges that limit their use as decision-support tools. A significant issue is parameter estimation in large-scale ABMs, particularly due to computational constraints on exploring the parameter space. This study evaluates a state-of-the-art simulation-based inference (SBI) framework that uses neural networks (NN) for parameter estimation. This framework is applied to an established labour market ABM based on job transition networks. The ABM is initiated with synthetic datasets and the real U.S. labour market. Next, we compare the effectiveness of summary statistics derived from a list of statistical measures with that learned by an embedded NN. The results demonstrate that the NN-based approach recovers the original parameters when evaluating posterior distributions across various dataset scales and improves efficiency compared to traditional Bayesian methods.

Neural Network-Based Parameter Estimation of a Labour Market Agent-Based Model

TL;DR

A state-of-the-art simulation-based inference framework that uses neural networks (NN) for parameter estimation is evaluated and it is demonstrated that the NN-based approach recovers the original parameters when evaluating posterior distributions across various dataset scales and improves efficiency compared to traditional Bayesian methods.

Abstract

Agent-based modelling (ABM) is a widespread approach to simulate complex systems. Advancements in computational processing and storage have facilitated the adoption of ABMs across many fields; however, ABMs face challenges that limit their use as decision-support tools. A significant issue is parameter estimation in large-scale ABMs, particularly due to computational constraints on exploring the parameter space. This study evaluates a state-of-the-art simulation-based inference (SBI) framework that uses neural networks (NN) for parameter estimation. This framework is applied to an established labour market ABM based on job transition networks. The ABM is initiated with synthetic datasets and the real U.S. labour market. Next, we compare the effectiveness of summary statistics derived from a list of statistical measures with that learned by an embedded NN. The results demonstrate that the NN-based approach recovers the original parameters when evaluating posterior distributions across various dataset scales and improves efficiency compared to traditional Bayesian methods.
Paper Structure (13 sections, 3 equations, 9 figures, 3 tables)

This paper contains 13 sections, 3 equations, 9 figures, 3 tables.

Table of Contents

  1. ABM parameter estimation
  2. Simulation-based inference for ABM parameter estimation
  3. SBI4ABM
  4. Labour Market Occupational Mobility and Automation ABM
  5. Methodology
  6. Applying the SBI4ABM pipeline in the Labour Market ABM
  7. LM-ABM economic indicators
  8. Dimensionality reduction
  9. SBI4ABM scalability
  10. U.S. labour market
  11. Validation of Bayesian inference with SBC
  12. LM-ABM as job transitions
  13. Conclusion

Figures (9)

  • Figure 1: Sampling and pairwise marginals from an NPE trained with a list of statistic measures. Labour market with 10 synthetic occupations
  • Figure 2: Sampling and pairwise marginals from an NPE trained with a NN-learned summary statistics. Labour market with 10 synthetic occupations
  • Figure 3: Average SBI4ABM simulation time per number of occupations
  • Figure 4: Average SBI4ABM training time per number of occupations
  • Figure 5: Sampling and pairwise marginals by learned summary statistics of the U.S. labour market
  • ...and 4 more figures