The Mass Spectrometry society of Japan - The 68th Annual Conference on Mass Spectrometry, Japan
Japanese

Abstract

Table of contents

Timetable

Plenary Lectures

Oral Sessions (Day1, Day2, Day3)

Poster Presentations (Day1, Day2, Day3)

Workshops

Luncheon Seminars (Day1, Day2, Day3)

Poster Presentations

Day 3, May 13（Wed.）　　Poster (1008/09)

3P-30　　PDF

Search for applications of mass spectrometry using pre-trained model

(1Shimadzu, 2Shimadzu)
oKentaro Morimoto1, Yuki Ishikawa1, Kanako Mita1, Akiyoshi Ohtani1, Yoshihiro Hayakawa2, Shigeki Kajihara1

We have developed search method for mass spectrometry application using neural language processing. This method enables us to search similar application to target sample based on context. One of the most rapidly development in the field of NLP was BERT(Bidirectional Encoder Representations from Transformers). It's built on pre-training contextual representations called pre-trained model. We adapted BERT pre-trained model for our search method.