Japanese
English
- 有料閲覧
- Abstract 文献概要
- 1ページ目 Look Inside
- 参考文献 Reference
人工知能による画像分類の画像診断における応用は,世界中で大変な勢いで開発が進んでおり,すでにいくつかの用途で商品化もされている。しかしながら,画像分類の精度という点では,まだ発展途上であり,ネットワークの構造や,最適化の方法も日々新しいものが提案されているというのが現状である。
This article introduces the two novel mathematical methods for diagnostics. The first one is the Capsule network, initially proposed by Hinton, which can learn positional information in images. We attempted to improve classification accuracy by replacing the classical convolutional network with the capsule network. Our method was able to achieve a higher classification accuracy than that of previous research. The next one is the persistent homology. It is a primary tool for topological data analysis, which is a recently advanced concept in applied mathematics. This article describes our experience with tumor classification.
Copyright © 2019, KANEHARA SHUPPAN Co.LTD. All rights reserved.