Swarm intelligence is an important branch of nature inspired algorithms where collective intelligence of different species like ants, bees, cuckoos, etc. is simulated. Even though swarm intelligence algorithms mimic very different phenomena, they all start with a collection of random points in the space of potential solutions and then search for better solutions, keeping balance between exploitation and exploration. We have recently successfully modified and improved several swarm intelligence algorithms. Here we present some of the techniques that we used for these modifications and improvements.