[1L4-OS-4a] OS-4

[1L4-OS-4a-04] Ensemble model visualization for MLOps based on relations among weak estimators

〇Masakazu Hirokawa¹, Miyu Kashiyama², Ryuta Matsuno¹, Keita Sakuma¹, Yoshio Kameda¹, Takayuki Itoh² (1. NEC Corporation, 2. Ochanomizu University)

Keywords:Visualization, MLOps

Machine learning methods, exemplified by deep learning and large language models, are becoming increasingly complex black boxes. This leads challenges not only in operational aspects like model maintenance and quality assurance but also in addressing societal needs such as fairness and privacy. Ensemble learning that combines multiple weak learners for enhanced performance, is widely used but suffers from low interpretability/explainability. In this paper, we propose a new visualization method focusing on the relationships between weak learners in ensemble models to improve understanding of model structures and learning processes. We demonstrate its utility through visualization of gradient boosting decision trees and discuss its potential for addressing operational challenges and supporting knowledge discovery tasks such as bias detection and drift analysis in datasets.