We consider in a new light the age old question of
discriminating signal and noise from finite samples. Applications in
mind are as diverse as radar, sonar, wireless communications,
bio-informatics and machine learning. We provide an
application-independent approach that brings into sharp focus a
fundamental finite-sample statistical limit of signal detection. What
emerges is yet another example of how random matrix theory is
transforming the theory and practice of statistical signal processing.
Continuing on this success, we highlight the random matrix origin of
several other related phase transition phenomena.
|