February 3, 2025
The Complex Event Recognition (CER) team, part of the AI Lab of the Institute of Informatics & Telecommunications, has been awarded a significant new patent (EP3955176A1) for an innovative system that redefines real-time event forecasting. This cutting-edge approach combines symbolic automata and prediction suffix trees, setting a new standard for accuracy, efficiency, and scalability in the analysis of streaming data.
The patented system employs symbolic automata to model complex event patterns while integrating variable-order Markov models for probabilistic forecasting. By focusing on the most informative past sequences, it captures long-term dependencies in data with exceptional precision, overcoming the limitations of traditional methods that often require extensive computational resources or domain-specific knowledge.
Key advantages include:
This system seamlessly constructs a probabilistic model from a user-defined pattern without requiring extensive domain knowledge, making it a versatile and highly adaptable tool for anyone working with streaming data.