Primary Submission Category: Causal Discovery
Intuition AI – A Hypotheses Iteration based Scientific Learning Framework
Authors: Rabindra Chakraborty,
Presenting Author: Rabindra Chakraborty*
For rare events of high consequence that exhibit extreme behaviors in biological, chemical, and geological systems, standard machine learning fails to perform with any meaningful accuracy due to lack of data. For all these occasions, esp., when ground truth is not available instantly, industries fall back on experts’ interpretations to avoid high stake consequences in an operation. Intuition Technology is a patented causation AI that builds scientific models using experts’ hypotheses that are then iterated towards situational ground truth using multi-view convergence, as various diverse situational datasets are run through the model. This makes builds the fabrication of a strong model despite lack of data.
The internal state of a natural system is not observable in most cases yet is often responsible for its unexplained behavior. Intuition AI is a framework that captures different situations, time delay between a cause and an effect and the degree of the effect viz-a-vis the influencers along with the experts’ explanations, thus building a situational response map useful for forewarning complex system behavior.
Intuition AI has been successfully piloted with two oil supermajors, one for interpreting real-time contamination state in fluid samples in wireline operations and the other for estimating reservoir characteristics using mud-gas logging and drilling data.