The concept of machine learning

The purpose of a service like Ai-Metrix® is to save time and to insure uniform, high-quality assessments of a chemical process. Since the processing is automated, there are decisions that must be made in software ahead of time that govern the model quality. These decisions fall into two categories, either qualitative or quantitative analysis. In recent literature, this is bundled into the term Machine Learning.

If you are interested in an independent overview of this topic, here is an interesting site with videos and a book you can download: https://www.r-bloggers.com/in-depth-introduction-to-machine-learning-in-15-hours-of-expert-videos/

Modularity

Infometrix has incorporated algorithms that, in our decades of experience, will best model the process being monitored by the spectrometer. The elements we have in place are separable modules, so when alternate technologies become available that outperform the originals, they can be swapped. This assumes the new approach is compatible with any software currently deployed for routine processing of spectral data. If your current modeling software handles your application, we can duplicate that process within the Ai-Metrix system.

Off-the-shelf components

As much as possible, Infometrix has incorporated well-established components to form the Ai-Metrix system. The database core is the industry standard Microsoft SQL Server, reports are presented in Adobe PDF, dashboards are presented via Tableau On-Line, and model quality is verified using Infometrix Pirouette.

Customizable settings

Infometrix scientists review the existing models for the spectrometer to establish processing guidelines for the automated process. In this way, the system mimics the current user as a starting point for model optimization. These settings can be reviewed at any time.

Help whenever you need it

Infometrix is always available to assist in understanding the results, strengths and limitations of the data and the models under management. We can review and further optimize the models, provide additional tools for in-house use, and supply training on the modeling process itself or on multivariate analysis in general. Because of the database structure, Infometrix reviews are typically straightforward and fast, requiring only a few days to complete a comprehensive review process.