Detecting Hidden Effects in High Dimensions
</br>
</br>
In this talk, I will discuss two problems. The first concerns the sample size that one would need in order for a simple epidemiological study of the association of genes and common diseases to produce meaningful results. This will be achieved by studying the power of the 2-sample Binomial test.
</br>
</br>
The second topic concerns the detection of a noisy signal hidden in a mass of random noise. I will discuss what happens if independent draws from a mixture density <MATH>(1-p) f(x) + p f(x-&mu;) (&mu;>0)</MATH> are observed and one would like to identify the observations coming from the second component of the mixture.
</br>
</br>
<a href="http://www.stat.columbia.edu/~tyler/seminar" title="Schedule">Back to the schedule</a>