TfELM: Extreme Learning Machines framework with Python and TensorFlow

Abstract

TfELM introduces an innovative Python framework leveraging TensorFlow for Extreme Learning Machines (ELMs), offering a comprehensive suite for diverse machine learning (ML) tasks. Existing solutions in the ELM landscape lack comprehensive implementations. TfELM fills this gap by consolidating 18 ELM variants (including 14 so-far unimplemented in Python) into a unified framework. It conforms to established scikit-learn standards and emphasizes modularity, facilitating seamless integration into ML pipelines. Harnessing TensorFlow’s GPU acceleration, TfELM ensures rapid training and compatibility across varied computing environments. Notably, TfELM marks the inaugural ELM implementation in TensorFlow 2, featuring high-performance model saving/loading via HDF5 format, thus enhancing its novelty and alignment with contemporary standards. Performance evaluations demonstrate that TfELM outperforms other solutions, achieving significant speed enhancements across various computing platforms, with improvements of up to nine times tested on five standard UCI datasets.

Keywords

• Extreme Learning Machine
• Machine learning
• Artificial intelligence
• Neural networks
• Python
• TensorFlow