랜덤 포레스트를 이용한 스마트폰 사용자의 행위 기반 분류의 글로벌 해석을 위한 시각화
Visualization for Global Interpretation of Activity Based Classification of Smartphone Users using Random Forest
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
모바일 어플리케이션에서 편리하고 지속적이면서 에너지 효율이 높은 사용자 분류 기술의 중요성이 증가하고 있으며, 다양한 사용자 움직임을 기반으로 성능이 좋은 사용자 분류 방법에 대한 연구가 활발히 이루어지고 있다. 하지만 현재 연구되고 있는 사용자 분류 기법은 사용한 기계 학습 모델이 크고 복잡하여 분류 결과를 해석하여 설명하기는 어려운 실정이므로, 이를 개선하여 연구자가 사용자 분류 결과에서 유사한 사용자와 아닌 사용자의 원인을 설명할 수 있는 방안에 대한 연구가 필요하다. 이를 위하여 본 논문에서는 하이퍼파라미터를 최적화하여 개발한 랜덤 포레스트 분류 모델을 적용하여 사용자를 분류하고, 그 결과에서 분류결정을 하게 된 글로벌 해석을 시각화하여 설명으로 제공하고자 설명가능 AI(XAI)의 SHAP(Shapley addictive explanation)을 사용하여 제시한다. 실험을 통하여 실험 데이터에 있는 특징 중, 어떤 특징들이 사용자 분류에 가장 크게 기여하는 것인지 요약한 결과 및 정분류된 사용자와 오분류된 사용자의 인스턴스를 해석한 결과를 제시하였다. 그 결과, 랜덤 포레스트 분류 모델이 사용자 분류 과정에서 사용자를 분류한 이유를해석할 수 있음을 보였으며, 사용자 행위 센서 데이터를 사용한 사용자 분류의 설명 가능한 기법으로 SHAP 을 사용할 수 있음을 제시하였다.
Abstract
The importance of convenient, continuous and energy-efficient user classification technology is increasing in mobile applications, and research on user classification methods with good performance based on various user movements is being actively conducted. However, the user classification techniques currently being studied have a large and complex machine learning model, so it is difficult to interpret and explain the classification result. So research is needed to study how to improve this so that researchers can explain the causes of similar users and dissimilar users in the user classification results. In this paper, we classify users by applying the Random Forest classification model developed by optimizing the hyperparameters, and use SHAP(Shapley addictive explanation) of explainable AI(XAI) to visualize and provide the global interpretation for the classification decision from the results. Through the experiment, the results of summarizing which of the features in the experimental data contributed the most and the results of interpreting instances of correctly and misclassified users were presented. As a result, it was shown that the Random Forest classification model can interpret the reason for the classification based on the result, and suggested that SHAP can be used as an explainable technique for the user classification using user activity sensor data.
참고문헌 (23)
- Mohammed Abuhamad, "Sensor-Based Continuous Authentication of Smartphones’ Users Using Behavioral Biometrics: A Contemporary Survey", IEEE Internet of Things Journal, 8(1), pp.65~84 (2021), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- Ahmed Fraz Baig, "Security, Privacy, and Usability in Continuous Authentication: A Survey", Sensors, 21(17), pp.5967~ (2021), MDPI AG[DOI]
- Yunji Liang, "Behavioral Biometrics for Continuous Authentication in the Internet-of-Things Era: An Artificial Intelligence Perspective", IEEE Internet of Things Journal, 7(9), pp.9128~9143 (2020), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- Muhammad Ehatisham-ul-Haq, "Authentication of Smartphone Users Based on Activity Recognition and Mobile Sensing", Sensors, 17(9), pp.2043~ (2017), MDPI AG[DOI]
- K. Chen, "Deep learning for sensor-based human activity recognition:overview, challenges and opportunities", arXiv preprint arXiv:2001.07416, ~ (2020)
- Atul Rawal, "Recent Advances in Trustworthy Explainable Artificial Intelligence: Status, Challenges, and Perspectives", IEEE Transactions on Artificial Intelligence, 3(6), pp.852~866 (2022), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- I. Kok, "Explainable artificial intelligence (xai) for internet of things: a survey", arXiv preprint arXiv:2206.04800, ~ (2022)
- Alejandro Barredo Arrieta, "Explainable Artificial Intelligence (XAI): Concepts, taxonomies, opportunities and challenges toward responsible AI", Information Fusion, 58, pp.82~115 (2020), Elsevier BV[DOI]
- Leo Breiman, "Random forests", Machine Learning, 45(1), pp.5~32 (2001), Springer Science and Business Media LLC[DOI]
- Zifeng Wu, "A Comprehensive Study on Cross-View Gait Based Human Identification with Deep CNNs", IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(2), pp.209~226 (2017), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- Weizhi Meng, "Surveying the Development of Biometric User Authentication on Mobile Phones", IEEE Communications Surveys & Tutorials, 17(3), pp.1268~1293 (2015), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- Maria De Marsico, "A Survey on Gait Recognition via Wearable Sensors", ACM Computing Surveys, 52(4), pp.1~39 (2019), Association for Computing Machinery (ACM)[DOI]
- Muhammad Muaaz, "Smartphone-Based Gait Recognition: From Authentication to Imitation", IEEE Transactions on Mobile Computing, 16(11), pp.3209~3221 (2017), Institute of Electrical and Electronics Engineers (IEEE)[DOI]
- Nicholas D. Lane, "An Early Resource Characterization of Deep Learning on Wearables, Smartphones and Internet-of-Things Devices", Proceedings of the 2015 International Workshop on Internet of Things towards Applications, pp.7~12 (2015), ACM
- Tin Kam Ho, "Random decision forests", Proceedings of 3rd International Conference on Document Analysis and Recognition, pp.278~282 (1995), IEEE Comput. Soc. Press
- Rung-Ching Chen, "Selecting critical features for data classification based on machine learning methods", Journal of Big Data, 7(1) (2020), Springer Science and Business Media LLC[DOI]
- S. M. Lundberg, "A unified approach to interpreting model predictions", Advances in neural information processing systems, 30 (2017)[DOI]
- M Lundberg Scott, "Explainable ai for trees: From local explanations to global understanding", arXiv preprint arXiv:1905.04610, ~ (2019)
- Scott M Lundberg, "Consistent individualized feature attribution for tree ensembles", arXiv preprint arXiv:1802.03888, ~ (2019)
- Scott M Lundberg, "Explainable ai for trees: From local explanations to global understanding", arXiv preprint arXiv:1905.04610, ~ (2019)
- D. Anguita, "A public domain dataset for human activity recognition using smartphones", European Symposium on Artificial Neural Networks, pp.437~442 (2013)
- F. Pedregosa, "Scikit-learn: Machine learning in Python", the Journal of machine Learning research, 12, pp.2825~2830 (2011)[DOI]
- "Explainable AI (XAI)with SHAP -Multi-Class Classification Problem" (2021)[URL]