Page 37 - RC19 RealcommEDGE 2019 Fall Issue

A three-pronged AI strategy of

objective categorization, ecosystem
learning, and simplified in-process
application can help transform

reactive processes for comfort and
efficiency into a proactive process.

For any model, the cost of creating a ‘learning set’ i.e., a into simplified metrics which can be easily interpreted. It
labeled training set and test set, is enormous. It is also an also shouldn’t completely change the existing process,
iterative process. Ecosystem learning is about creating a but improve it and make it more user-friendly. For exam-
system where learning happens through an ecosystem of ple, one could change a descriptive field for entering
use. Learning involves creation of a labeled dataset which failed parts to a category selection. This will help in the
signifies a particular objective to predict based on the past adoption of the ecosystem, which is critical for the suc-
data set. For example, when building a model for predict- cess of not only the evolution of the models, but also the
ing equipment failure, the existing data set needs to be transformation into a predictive, data driven-process.
labeled for confirmed failures. To have the model predict
individual components and reasons for failure, the labeling It’s time to adopt models that consider multiple per-
needs to be done to address these specific details. With spectives with the right insights in context, and gain
additional objectives, the complexity increases multifold significant benefits in compliance, health and efficiency
while the accuracy of the labels decreases. across buildings and locations. Diagnose the symptoms,
act proactively with the right tools and do not allow your
The ecosystem learning approach addresses this chal- building and equipment operations to deteriorate.
lenge by creating an ecosystem where the base objective
is applied first and then operationalized with users, who Mansoor Ahmad is the Global Business Head of
help create the labeled data set for the secondary objec- EcoEnergy Insights. He has been influencing the future
tives. In the above example, a combination of the details of Building IoT through EcoEnergy Insights’ disruptive
of the parts replaced, together with the confirmation that innovations. He is responsible for steering a start-up
the equipment health improved, would ascertain that incubated in 2009 within an IT Services company, into
indeed the equipment had failed due to the failure of the EcoEnergy Insights of today—a leading provider of digital solutions
part. This can confirm the secondary objective: predic- and services for optimizing building and equipment to some of the
tion of individual component failures. world’s largest enterprises.

Hence the design of the AI/ML model should also look Subhasis Mandal heads Technology, New Products
at the operationalization and the ecosystem to improve and Services Development at EcoEnergy Insights. He
without increasing the cost. has been instrumental in conceptualizing and
architecting data-enabled, platform-driven services
Another important strategy is the simplified in-process like energy efficiency, maintenance and associated
application of the model results. When an ecosystem is solutions. He was the key driving force behind the design and launch of
being created for users, the results need to be converted the CORTIX Building IoT platform.
™

35