Loading...
AI capabilities
Soft reasoning
Sub-symbolic reasoning systems use deep learning models and probabilistic techniques to process complex, ambiguous, or unstructured information that eludes strict logical formalisation. These systems, based on neural networks and large-scale language models, are capable of identifying subtle patterns, understanding complex semantic contexts, and generating probabilistic predictions from heterogeneous data. Sub-symbolic reasoning complements deterministic reasoning by addressing scenarios where uncertainty, linguistic ambiguity, or the unstructured nature of data requires statistical approximations and generalisation capabilities. This integration enables the development of hybrid solutions that combine the precision and traceability of logical reasoning with the flexibility and adaptability of machine learning, delivering more robust and versatile AI systems.
