JIITA, vol.6 no.2, p.557-566, 2022, DOI: 10.22664/ISITA.2021.6.2.557
Hyeonji Kim 1), Yoosoo Oh 2,*)
1,2) Dept.of ICT Convergence, Dept. of AI, Daegu University, Gyeongsan-si, Republic of Korea
Abstract: In machine learning and deep learning models, accuracy is determined by preprocessing of data. So the data preprocessing process is critical to learning in machine learning and deep learning. Therefore, in the deep learning model learning data analysis process, the data preprocessing process must be carried out to use the data collected by users. To proceed with the data preprocessing procedure, the user must analyze the collected data directly. This paper proposes an algorithm that identifies the learning data type collected by users. In addition, to make meaningful data, users must repeat the task of finding the appropriate preprocessing of the files they want to use. Therefore, this paper proposes a data preprocessing method recommendation algorithm for deep learning model learning to minimize cumbersome tasks. Recommendation algorithms are recommended by knowledge-based filtering. This paper verifies the performance of the recommendation algorithm through the evaluation of the Titanic classification model.
Keywords: Preprocessing data, Knowledge base filtering, Recommendation, data, deep learning
Fullpaper: