Recent Advances in Convolutional Autoencoder with Dual-Key Transformer Network-Based Causality Analysis of Human Resource Practices on Firm Performance: A Systematic Review

Abstract

In recent years, the integration of deep learning architectures into human resource analytics has transformed the way organizations evaluate the impact of HR practices on firm performance. This paper presents a systematic review of recent advances in convolutional autoencoders combined with dual-key transformer networks for causality analysis in HR systems. Convolutional autoencoders enable efficient feature extraction from high-dimensional employee data, while transformer-based architectures enhance the modeling of complex temporal and contextual relationships. The dual-key mechanism further introduces a secure and interpretable approach to causal inference by separating feature representation and relational learning. This study synthesizes current methodologies, datasets, and evaluation metrics used in this interdisciplinary domain. It highlights how these hybrid models improve predictive accuracy, scalability, and interpretability in HR decision-making processes. Additionally, the review identifies key challenges such as data privacy, model transparency, and limited availability of standardized HR datasets. The findings suggest that combining deep learning with causal inference frameworks provides a promising direction for optimizing workforce strategies and enhancing organizational performance. This paper serves as a comprehensive reference for researchers and practitioners aiming to leverage advanced AI techniques in human resource management.