b'THE SOLENIS SOLUTION Unlikenetworks, reveal multi-dimensionalchanging machine conditions. Figure OPTIX Applied Intelligence is a novelother datarelationships between the response2 shows how a static predictive adaptive analytics platform built withanalytic tools,variable and predictors that are notmodel can lose prediction accuracy the latest artificial intelligence (AI)OPTIXeasily identified other ways. Figureover time, whereas the machine and machine learning capabilitiesprovides a1 illustrates the data relationshiplearning algorithms built into OPTIX available today. OPTIX was developedreal-time,awareness without OPTIX asconstantly tune predictive models to provide an advanced digitalcalculatedcompared to the complex relationshipsand preserve prediction accuracy service for the mill of the future. Itvalue forthat can be identified using OPTIX.in real time. Each OPTIX model is incorporates Solenis customer- andcritical qualitySolenis employs a proprietaryalso reinforced with a self-diagnostic quality-first philosophy, which meansparameters. screening methodology that allowsprediction accuracy monitor, which it can be tailored to every millsOPTIX to focus on the most importantcalculates the difference between the unique processes. Using processtags needed to drive the machine- prediction and value of the actual lab data, laboratory measurements andlearning, predictive platform. measurement.tissue making process knowledge, Solenis applies robust data scienceIt is crucial for analytics platformsINFORMED DECISION-MAKINGto provide an adaptive soft sensor.to adapt to the ever-changingThe real-time, data-driven, quality Unlike other data analytic tools, OPTIXenvironment of the tissue makingmeasures delivered by OPTIX does not require time-consuming dataprocess. Solenis predictive modelsallow for more informed, on-the-interpretation but rather provides aincorporating machine learningfly decision making and enable a real-time, calculated value for criticaladapt to changes in the processstep change improvement in mill quality parameters. by learning from observations andoptimization not previously feasible. interactions. This information drivesOperators can confidently make Solenis data scientists carefullyshifts in the predictive model aimedmore refined, economical, and timely evaluate mill process data usingat maintaining prediction accuracy.process adjustments that are less sophisticated data collection, cleaningThis level of machine learning isprone to error or reactive to outliers. and mining techniques. Advancedbuilt into the platform, which isImproved process visibility and data mining practices, such aswhy OPTIX measurements remaininformed decision making provided multiple regression and causalrobust and accurate in the face ofby OPTIX enables mills to change Figure 2. (a) Predictive model using poor prediction techniques that oversimplify the process; (b) Predictive model with the same data utilizing OPTIX, which is capable of handling tissue machine process data using machine learning techniques TISSUE TECHNOLOGY INTERNATIONAL85'