Ahmed Hefny Photo

Ahmed Hefny

Machine Learning Department
Carnegie Mellon University

Office: GHC 8223
E-mail: ahefny AT CSdotCMU DOT edu
Home | Publications | Resume | Misc

My Way of Life

To see how I deal with the fundemantal questions in life like "Why are we here ?", "What is going to happen next ?", "Why am I doing this ?" and other questions every one is (supposedly) concerned about, check these links.

Computational Academics

These are my attempts to develop quantitative and rigorous (really ?) models to explain (and hence exploit) physical phenomena that occur in the academic environment

Lesson #1 - Uniform Bound

Many professors say the exam is going to be easy while they actually mean that each question is going to be easy. Don't be deceived by this. As we all know, if the difficulty of a question tends to 0, that does not necessarily imply that the difficulty of union of N questions will tend to 0 as N goes to infinity (which does happen !!!)

Lesson #2 - Correlation

Exam questions are NOT i.i.d. Exploit this fact !
Example:
Question 3: Derive the form of the likelihood ratio confidence interval for t for Uniform(0, t)
Question 4: Show that [X(n),X(n)/a^(1/n)] is a confidence interval for t for Uniform (0, t)

Gaming

Since childhood, I was assumed to be a hard-core gamer (it is not a coincidence that my first published work was on machine learning for games). However, compared to insane skills over the internet, I classify myself as a rather moderate player, though I don't currently have enough time (or hardware) for this. But just in case, if you find yourself playing against cyber_modiko somewhere, please do have mercy !
Last update: Feb 16, 2013