I have been reading the book Multidimensional Particle Swarm Optimization for Machine Learning and Pattern Recognition. They use different norm for different problems in their automatic ECG detection system. Therefore, they select a general class of norms l2 Minkowski norm, but they do not use Poincare group, instead Euclidean group.
The book does not specify exact groups for the system. The broad variability of norms suggests me a weakness in the selection of groups. They do not also specify the TFR, but I believe some of it is based on Wigner-Ville distribution with one integral.
How do you need different norms for different problems in automatic detection system of ECG?
I got an answer from my professor that you can use Minkowskian space without the selection of specific group like Poincare.