|
|
|
|
Safety_Critical_Machine_Learning |
|
|
|
|
|
AI algorithms are capable
of gathering and processing data from numerous different sources to aid
decision making. In many circumstances AI systems can analyze a situation
quickly and efficiently and make the best decision in a critical situation.
As ML-based AI learns from historical data there is a danger that it will
learn from biases that may exist in the database used for learning. Although
the algorithm could neutralize the biases that derive from human input, there
are limitations of AI as a decision maker on an ethical and human level.
However, in decision making under pressure AI algorithms can play an
essential role in cooperation with humans. For the combination of humans'
ethical understanding and artificial intelligence's rapid analytical
capabilities to accelerate decision-making, AI algorithms need to be
understandable. If, as in the case of Deep-Learning, the algorithm takes a
decision in an inexplicable way, there can be no effective human-algorithm
collaboration. S2-EX-AI-DED (Strategic
Scenario EXplainable AI Decision Expert Doer) has
been realized as a POC application to test the explainability
of SHARP™ in a Command and Control context. The application of the SHARP™
neural network enables the user to recover the examples that determined the
inference: this feature is fundamental for algorithm-human decision-making
collaboration. |
|
|
|
|
|
|
General Synaptics Aerospace_and_Defence_Machine_Learning_Company VAT NUMBER:_IT02670700992 REA NUMBER: GE-503104 Email:_luca.marchese@synaptics.org |
|
|
|
Copyright© 2026 General Synaptics |
