Event time:
Monday, November 6, 2017 - 3:50pm to 5:00pm
Location:
LOM 206
Speaker:
Nir Sharon
Speaker affiliation:
Princeton
Event description:
Multireference alignment is the problem of estimating a signal from its circularly translated copies in the presence of noise. We show that in the regime of low SNR, the distribution of translations plays a significant role in terms of the required number of samples. We use lower-order statistics for solving the problem, establish some theoretical guarantees, and introduce practical methods.