Hybrid AI

In recent years, AI based on deep learning has achieved tremendous success in specialized tasks such as speech recognition, machine translation, and the detection of tumours in medical images. Despite these successes there are also some clear signs of the limitations of the current state-of-the-art in AI. For example, biases in AI-enabled face recognition and predictive policing have shown that prejudice in AI systems is a real problem that must be solved. In this position paper, we argue that current AI needs to be enhanced along four dimensions to become more trustworthy: environment, purpose, collaboration, and governance. Hybrid AI offers the potential for advancements along these four dimensions by combining two different paradigms in AI: knowledge-based reasoning and optimization, and data-driven machine learning. Some hybrid AI design patterns show how these paradigms can be combined to harness the advantages of both approaches while at the same time overcoming their limitations. We introduce two classes of systems that are enabled by hybrid AI: autonomous systems and human-machine teams. Several examples show how hybrid AI can be employed to make these system classes more trustworthy. Hybrid AI technology is developed and applied in almost all Appl.AI projects.

What does TNO offer on Hybrid AI?

TNO develops hybrid AI architectures and algorithms for application domains such as autonomous vehicles, healthcare, and safety & security. For autonomous robots, TNO has developed a hybrid AI architecture that uses a central embedded knowledge base to improve the situational awareness and self-awareness that can be provided by machine learning algorithms. For skills matching, i.e. the matching of job descriptions with CVs of job seekers, a skills ontology and a neural network are combined to learn which sentences or parts of sentences of a vacancy text contain a skill that matches a specific skill in the ontology.

In cooperation with the Free University in Amsterdam, TNO is developing a taxonomically organised vocabulary and modular design patterns to describe both processes and data structures used in hybrid AI.