ICCV Done

Is the PhD life all about deadline?

Well, I am such a guy who always more willing to look at the bright side. Even the paper doesn't make it finally, still we learned a lot of things during each process. Discussion, implementation and struggling about the way to present the idea.

Maybe we can learn like 50% through class about one specific method, but after writing a paper using that method, we can leverage it to say 80%. This time we use design a EM algorithm intuitively, but how to formalize it into the right way cost more time than implementing the method.

Where is the data, and what is the parameter and what is the indicator? What does the E step do, how can we get there? And what does the M step do, the deducing path is correct or not? Still, we are very lucky that Hal helped us get through this painful process....

It seems to be a normal routine. We usually design the method first, and can't wait to implement it to see whether it gets better result or does what we expected. Then if yes, we sit down and try to figure out how to formalize what we've already did in a formal way...


y