ford and fulkerson.

i remember that when i was taking the operational research course at the ppm (powerful politecnico di milano) i especially liked the ford-fulkerson algorithm (and i disliked the simplex algorithm).
i liked the ford-fulkerson algorithm because it was easy and it was about computing how much water you can possibly pump through a network (one of my grandfathers is a plumber, so i like water pipes, etc.).
it never seemed very useful or brilliant to me, though. so i forgot about it quite fast.

then, some years later, i was attending a tutorial by phil torr at the bmvc and it turned out that solving that water distribution problem is equivalent to minimizing a cost function, so you can apply it to bayesian image restoration and image segmentation and a lot of other cool stuff!
i was amazed.

well done, ford and fulkerson! ;-)


i have some more stuff i should write about my trip to leeds, i'll write it when i have more free time, maybe on the week end.

1 comment:

Giovanni Prestige said...

Bring on the updates! Thanks.