Public Seminar on Property Prediction Learning via Near-Infrared Spectrum Analysis to be Held on June 2nd

IR Technical Education Research Institute will hold a public seminar that explains the concept and procedures for real-time prediction of characteristic values and content, which are essential for quality control, by utilizing near-infrared spectra, interspersed with industrial examples from agriculture, food, and pharmaceutical manufacturing. In manufacturing and production sites, it is crucial to "quickly grasp the characteristics and content of objects" for stabilizing quality and improving yield. Traditionally, obtaining characteristic values often involved sampling and chemical analysis, which tended to be burdensome in terms of time and cost. To address these challenges, the utilization of prediction using near-infrared spectra (soft sensors) is expected to enable real-time monitoring and advanced process control.

Seminar Overview

• Seminar Name: Property Prediction and Quantitative Model Construction by Near-Infrared Spectrum Analysis

• Format: On-site attendance / Live streaming attendance (Zoom)

• Date & Time: [On-site attendance] [Live streaming attendance] June 2, 2026 (Tuesday) 13:30-17:00 (including break)

• Location:

  [On-site attendance] Japan IR Co., Ltd. Head Office Seminar Room (3 minutes walk from Akihabara Station, etc.)

  [Live streaming attendance]: Online attendance (*Online lecture via Zoom)

• Capacity: 16 people for seminar room, unlimited for online

• Fee: 39,600 yen (tax included) / 1 person (group discount available)

• Instructor: Professor Kimihito Funatsu (Director / Specially Appointed Professor, Data-Driven Science Creation Center, Nara Institute of Science and Technology)

Course Details

This course covers the basics to applications for predicting concentration and characteristic values using near-infrared spectra. As a soft sensor for process monitoring, it will clarify "what can be understood" from near-infrared spectra, then teach practical essentials such as noise processing, preprocessing, quantitative model construction, wavelength selection, and region selection. Furthermore, it will introduce examples such as rapid estimation of soil properties, real-time internal quality prediction of fruits (application to automatic sorting machines), and component content monitoring in continuous pharmaceutical processes, expanding the discussion to property prediction and content quantification in chemical product manufacturing.

Seminar Program

1. What is Process Informatics?
2. What to make leads to how to make it
3. Soft sensors for process monitoring
4. What can be understood from near-infrared spectra
5. Noise processing of near-infrared spectra
6. Construction of quantitative models for concentration and properties using near-infrared spectra
7. Wavelength selection and construction of interpretable quality models
　7-1 Genetic Algorithm (GA) and wavelength selection / 7-2 Quality evaluation of tea leaves
8. Wavelength region selection (GAWLS)
　8-1 Application to rapid estimation of various soil properties
　8-2 Real-time prediction of internal quality of fruits (application to automatic sorting machines)
9. Quantitative models that do not use the form y=f(x)
　9-1 IOT / 9-2 IOT + GAWLS / 9-3 Application examples to continuous pharmaceutical processes
10. Application to property prediction and content quantification of chemical products
11. Vision for future soft sensors and process informatics

Target Audience

• Engineers engaged in process monitoring

• Researchers engaged in material research and development

• Process designers and technical team leaders involved in practical work