Artificial Intuition – Best Application Scenarios

Over the past twenty years, Ontonix has explored and worked on hundreds of applications of its QCM-powered Artificial Intuition technology, in many cases in industrial projects with large corporations. This blog focuses not on listing these application cases, or categorizing them. Rather, the goal is to show what conditions and contexts can motivate the usage of this unique technology regardless of market verticals. Artificial Intuition is application independent. However, there are cases in which the technology shines. Here is the list:

Conditions of high uncertainty and variability

Situations in which the dynamics of a particular system or process are dominated by uncertainty can be particularly nasty, especially if the underlying physical phenomena contain discontinuities, bifurcations, non-linearities, clustering, transitions, etc. This makes model building or model training and validation very difficult.

Data scarcity

Many anomalies or malfunctions which affect complex systems – both natural and man made – are often very rare and/or unique. There are simply not enough examples for a Machine Learning approach, no patterns to recognise. And yet, a decision must be made, often on the fly.

Prioritization

Prioritization relies on identifying the relative importance or order of variables, tasks, or goals. It’s about finding out what matters most. In complex situations this can be very difficult. In any situation, complex or not, knowing where to focus and when saves time and resources.

The need for explainability

It is known that AI is in many cases a black box that produces an answer but without offering explanation as to how the answer has been arrived at. This is a major shortcoming of AI. Being able to justify, for example, why an anomaly has been spotted, is key towards really understanding how things work and why and how they break. Artificial Intuition offers 100% explainability because it doesn’t guess the answer, it computes it.

Early anomaly detection

Artificial Intuition is particularly good at spotting the onset of anomalies and malfunctions, providing early precursors and indicating the underlying causes. This can be accomplished in real time, monitoring sensor outputs to provide instantaneous indications of potential problems before they materialize in a threatening manner. Artificial Intuition doesn’t require training. This makes it very fast and easy to implement in the context of edge computing, onboard any mobile platform.

Autopsy of a collapse

When highly complex systems fail they can do so in very many, often unexpected, non intuitive and unique ways. Sometimes the causes of a collapse are never determined or fully understood even though abundant data may be available for a post-mortem analysis. Complex catastrophic collapses offer no opportunity of any form of training, ruling out any application of AI-related techniques.

Acceleration of Machine Learning

One of the numerous applications of Artificial Intuition is that of accelerating Machine Learning, a long and energy-intensive process, requiring expensive computational hardware and large amounts of training data. Artificial Intuition can eliminate variables that do not contribute information, hence reducing the size of the training set.

To find out more about the numerous applications of Artificial Intuition visit our blog.