The power of complementarity

Technology is complementary to human skills, not competing, as humans and machines are good at different things

Machine learning tools excel at efficient data processing, where big amount of data are computed for a specific purpose.

Humans are better at making judgements based on complex data

• Therefore, computers do not replace experts but allow them to do more by:
  • Helping to automate repetitive tasks, such as:
    • Automating image analysis to increase efficiency and reproducibility of the process
  • Analysing huge amount of high-dimensional, multimodal, unstructured data, such as:
    • Fault detection in continuous process
    • Product quality
    • Defect prognosis
  • Reduce the number of options to be tested experimentally by ‘pre testing’ using in silico models, such as:
    • Chemical properties prediction before the substance is synthetized in the lab by a chemist
    • Prediction of biological effect (e.g. toxicity) of big number of substances before performing in vivo experiment with smaller, preselected group
• Machine Learning provides accurate insights for better decision making, applicable for processes such as (bio)production, operations, fault detection, maintenance, decision support, product quality improvement

Questions which can be answered with machine learning:

DESCRIPTIVE: What happened?

E.g. process, product quality

DIAGNOSTIC: Why it happened?

E.g. fault assessment, monitoring of equipment conditions or chemical process

PREDICTIVE: What will happen?

E.g. defect prognosis, yield prediction

PRESCRIPTIVE: What action to take?

E.g. adjusting parameters