BAI-OPTIMA
Major active project
BenAI OPTimal Individualised decision MAking foundational framework and algorithms based on future intelligence envisaged by BenAI
BAI-OPTIMA originated from Clinical Decision Support System and has the ambition to be a common decision making foundational framework for critical areas provided appropriate reasoning architecture found and generalised and domain knowledge imported.
Major Innovations
Maths
Decision logic (D-logic) extends traditional classification based decision-making
Architecture
- The GID model: a prediction module + rules module + a utility function + an rule-enforcement-module
- Enabling small dataset training and limited generalisation by introducing rules and partial knowledge
Algorithms
- The logic-MCUR algorithm: logic plus Multiple Criteria decision analysis within Uncertain Range, implementation of D-logic
- The DONDE algorithm: model-agnostic AI explainability
- The DEENA algorithm: extract domain data/rules from domain corpus described in natural language
- The SIR algorithm: find similar cases
More details in BAI-OPTIMA.pdf