Causation-based machine learning for early and accurate assessment of a situation.
Intuition AI was developed with a data-light approach, making deterministic predictions with as few as five data sets.
Our software utilizes the knowledge of domain experts to build dynamic and flexible models that understand situational dynamics that correlation-based models can’t discern.
The traditional “funnel” narrows relevant variables when training the model. Senslytics Intuition AI utilizes all influencers provided and dynamically selects those that are most relevant in a given situation.
Unlike traditional machine learning, the Senslytics model combines domain expert knowledge with dynamically selected data allowing for robust, deterministic conclusions.
When data is received by the Intuition AI platform, time series data follows one path while discrete events, situational data, and hypotheses follow a different path.
Time Series Data
Hypotheses and Situational Rule Management
Time Series Data
Time series data is divided into three tiers: Core, Ring, and Neutral. Each core influencer is considered a “view” and given a fidelity score that changes in real time.
Core Influencers provide the views used to determine if an event or state change is imminent. These influencers include information like sound speed and absorbance.
Ring Influencers impact core influencers and are always changing. This can include information such as pump rate.
Neutral Influencers are unknowns. These are the factors that limit current ML models from scaling when applied in new environments. Intuition AI uses these as a way to discover the unknown unknowns.
Hypotheses and Situational Rule Management
Discrete events, hypotheses, and situational data are used to determine which core influencers are most relevant in a given situation and time. Our patented situational rule runs a process of hypothesis iteration that tests and scores domain expert hypotheses. Those that prove to be true become situational rules, and those that don’t are removed from the process.
The event or state change being evaluated is added to a situational map where it is mapped to its nearest neighbor, along with Ring Influencers. This determines what situational rules and discrete events are relevant. Our patented dynamic selection of influencers works in a situational context to determine relevant core influencers.
Senslytics software can identify the cause of an event and validates data through multiple signatures. It then generates conclusions that are deterministic rather than probabilistic.
Our proprietary software finds signatures indicating when an event or state change is about to occur. When there is convergence of multiple Core Influencer signatures, our Intuition AI can indicate an inference or event with confidence.
Unlike most machine learning applications, Senslytics’ conclusions are deterministic, rather than probabilistic.
