Researchers have developed PaGNet, a novel hybrid model that combines Gradient Boosting Decision Trees (GBDT) with neural networks to forecast corporate tax avoidance proxies. This model addresses the challenge of extracting predictive signals from complex firm-year panel data while ensuring transparent model behavior. PaGNet utilizes a two-branch architecture, with one branch employing LightGBM for panel-temporal summaries and the other using a Panel-MLP with attention-based temporal aggregation and multi-task learning. Evaluations on Korean corporate data demonstrated PaGNet's ability to improve explained variance over existing baselines, particularly for accrual targets, and provide explicit branch-reliance reporting. AI
IMPACT Introduces a novel hybrid model for financial forecasting, potentially improving accuracy and transparency in corporate tax analysis.
RANK_REASON The cluster contains a research paper detailing a new model for forecasting. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
