Simple problems (max, min, sum, etc.)

  • Everybody knows how to do these

Complex problems (best result is Deterministic in Polynomial time):

  • Lots of classic algorithms out there, everybody knows how to find it on the Internet

Complex problems (best result is Non-deterministic in Polynomial time (NP problems)):

  • Greedy and Tactical Optimisations
  • Genetic: Find improvements by changing, mixing until no more improvements found
  • Perceptron/Neuron/Neural Network: Find improvements by changing weights until no more errors found on existing sample data entries.

Neural network for shared taxi pick-up/drop-off along the way:

  • Each car has its own neural network, the input & output below are for 1 car
  • Input: Coordinates of waiting passengers, each pair of coordinates is 1 input (to 1 dendrite)
  • Output: Coordinates of the passenger to visit first, this pair of coordinates is output (to axon terminals)
  • Three points above are just draft idea, may not really working!
Advertisements