[4Pin1] インタラクティブ(2)

[4Pin1-43] 事例紹介：24時間周期データに対する教師無し学習の適用

〇後藤 勲¹ (1. 住友電気工業株式会社)

キーワード：k-means、確率的潜在意味解析、ベイジアンネットワーク、24時間周期データ

In this article, I’d like to introduce three case studies for which I tackled for three years as a pilot study. This pilot study was applied to three fields. 1. Energy consumption in houses. 2. Anomaly detection of photovoltaic power system at Mega-solar system. 3. Sunlight combination type plant factory. In each case 24-hour period data was treated as a frame (but three data formats and types are different), and was analyzed in several methods. These analyses found out beneficial knowledge for each field.