Two-Stage Feature Generation with Transformer and Reinforcement Learning

2025/05/28 by Wanfu Gao, Zengyao Man, Gao, Wanfu +9
Computer Science · #Advanced Neural Network Applications #Domain Adaptation and Few-Shot Learning #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning and Data Classification

paper · pdf · doi:10.48550/arxiv.2505.21978

openalex publication_date 2025/05/28 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

Feature generation is a critical step in machine learning, aiming to enhance model performance by capturing complex relationships within the data and generating meaningful new features. Traditional feature generation methods heavily rely on domain expertise and manual intervention, making the process labor-intensive and challenging to adapt to different scenarios. Although automated feature generation techniques address these issues to some extent, they often face challenges such as feature redundancy, inefficiency in feature space exploration, and limited adaptability to diverse datasets and tasks. To address these problems, we propose a Two-Stage Feature Generation (TSFG) framework, which integrates a Transformer-based encoder-decoder architecture with Proximal Policy Optimization (PPO). The encoder-decoder model in TSFG leverages the Transformer's self-attention mechanism to efficiently represent and transform features, capturing complex dependencies within the data. PPO further enhances TSFG by dynamically adjusting the feature generation strategy based on task-specific feedback, optimizing the process for improved performance and adaptability. TSFG dynamically generates high-quality feature sets, significantly improving the predictive performance of machine learning models. Experimental results demonstrate that TSFG outperforms existing state-of-the-art methods in terms of feature quality and adaptability.

Citations

• OpenFE: Automated Feature Generation with Expert-level Performance
• Group-wise Reinforcement Feature Generation for Optimal and Explainable Representation Space Reconstruction
• A Survey on Deep Reinforcement Learning for Data Processing and Analytics
• DIFER: Differentiable Automated Feature Engineering
• Generative Partial Multi-View Clustering
• Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
• AutoCross: Automatic Feature Crossing for Tabular Data in Real-World Applications
• The autofeat Python Library for Automated Feature Engineering and\n Selection
• Feature Engineering for Predictive Modeling using Reinforcement Learning
• Proximal Policy Optimization Algorithms
• On Calibration of Modern Neural Networks
• Attention Is All You Need
• Sequence to Sequence Learning with Neural Networks
• Neural Machine Translation by Jointly Learning to Align and Translate
• Generating Sequences With Recurrent Neural Networks
• A few useful things to know about machine learning

Related

• Traceable Group-Wise Self-Optimizing Feature Transformation Learning: A Dual Optimization Perspective
• Traceable Automatic Feature Transformation via Cascading Actor-Critic Agents
• Feature Engineering for Predictive Modeling using Reinforcement Learning
• Dual-Agent Reinforcement Learning for Automated Feature Generation
• Self-Optimizing Feature Transformation