The main goal of the project is the development of a Computational Intelligence Platform for Evolving and Robust Predictive Systems (INFER) applicable in a number of different industries as well as its instantiation in a form of evolving, robust soft sensors for monitoring, diagnostic and control purposes in processing plants. This goal will be achieved by pursuing the following objectives within three overlapping research and partnership programme areas:

1. Area: Computational Intelligence – Objective 1:
Research and development of advanced mechanisms for adaptation, increased robustness and complexity management of the recently proposed open architecture for multi-component, multi-level evolving predictive systems.


  • Development of integrated adaptive/learning algorithms and approaches working on different time scales from real time adaptation to life long learning and optimisation.
  • Study and development of meta-learning approaches and techniques focusing on knowledge acquisition, representation and transfer for effective learning based on solving multiple problems.
  • Research on complex learning systems at the systems level with a focus on multi-component, multi-level, multi-objective optimisation and mechanisms (i.e. diversity, pool sizes, dynamics of adaptation, interactions, competition versus collaboration etc.)

Expected outcomes:

  • Development of the fundamental building blocks, algorithms and approaches to be used in achieving Objectives 2 and 3.
  • Publications in high quality journals and international conferences and additional dissemination through organisation of conferences, workshops and special sessions.
  • Contribution to the state of the art of computational intelligence, machine learning, complexity science, intelligent data analysis and smart adaptive systems fields with the promotion of a paradigm shift from human labour and knowledge intensive processes of building predictive systems to autonomous, evolving complex systems.

2. Area: Software Engineering – Objective 2:
Development of professionally coded INFER software platform for robust predictive systems building and intelligent data analysis.


  • Study and development of software engineering practices and methods in the context of pervasively adaptive complex software systems.
  • Generation of the system design for user interface for effective interaction with users and operators of complex engineering and commercial systems.
  • Description of the interactions between the users and continuously adapting INFER system.
  • Validation of INFER system within a real-world process industry environment in the context of Objective 3.

Expected outcomes/benefits:

  • Significant contribution to the state of the art of multi-component interacting software systems disseminated through high quality publications.
  • Development of commercial predictive systems software applicable in different industries and predictive modelling scenarios.
  • Creation of spin-off company for commercial exploitation of the R&D work carried out during the project.

3. Area: Process Industry / Control Engineering – Objective 3:
Development of self-adapting and monitoring soft sensors for process industry.

Challenges: The new generation of soft sensors should:

  • Identify their own validity and take appropriate actions to maintain and re-gain their validity if necessary;
  • Adapt and evolve in response to changing environment, to internal process dynamics and changes of the quality of the underlying data;
  • Re-design themselves or make the process of re-designing much more straightforward;
  • Allow efficient implementation of expert knowledge and provide effective channels for the user interaction at any level of information processing;
  • Be easy to install into new processes and demand minimum amount of maintenance;
  • Provide a certain degree of transparency in order to increase their acceptance in the industry.

Expected outcomes/benefits:

  • Scalability – the ability to deploy and maintain new soft sensors per plant at minimal additional cost.
  • Transferability – development of models that can be transferred to similar processes with minimum effort
  • Operational excellence – streamlined development of soft sensors for example for improved energy consumption, minimisation and monitoring of the pollutants level emitted into the atmosphere, improved product quality monitoring or reduction of the current requirement for manual, off-line product quality analysis to mention only few of the expected applications and improvements.
  • Large reduction of the soft sensor development and maintenance effort, thus making their application to small and medium size plants economically viable which in the majority of situations is not currently the case.

The main innovation of the project is a novel type of environment in which the ‘fittest’ predictive model for whatever purpose will emerge – either autonomously or by user high-level goal-related assistance and feedback. In this environment, the development of predictive systems will be supported by a variety of automation mechanisms, which will take away as much of the model development burden from the user as possible. Once applied, the predictive system will exploit any available feedback for its performance monitoring and adaptation.