11/20_3rd year seminar

Today, I introduced a paper titled "Analysis for Proposing Personalized Lifestyle Improvement Plans Using Bayesian Networks." This study uses Bayesian networks to infer the effects of lifestyle improvements on factors contributing to metabolic syndrome: hypertension, dyslipidemia, and abnormal glucose metabolism.

There are two types of Bayesian network models: data-driven models and knowledge-based models. According to the study, the knowledge-based model demonstrated approximately 20% higher accuracy. This result was quite surprising to me. While data-driven models often receive more attention, this study highlighted the potential of knowledge-based models as well.

On the other hand, the paper also mentioned that achieving high accuracy was difficult due to the insufficient amount of data. This reminded me of the critical importance of both the quantity and quality of data when conducting machine learning research.