Heterogeneous multireference alignment: A single pass approach

Abstract

Multireference alignment (MRA) is the problem of estimating a signal from many noisy and cyclically shifted copies of itself. In this paper, we consider an extension called heterogeneous MRA, where K signals must be estimated, and each observation comes from one of those signals, unknown to us. This is a simplified model for the heterogeneity problem notably arising in cryo-electron microscopy. We propose an algorithm which estimates the K signals without estimating either the shifts or the classes of the observations. It requires only one pass over the data and is based on low-order moments that are invariant under cyclic shifts. Given sufficiently many measurements, one can estimate these invariant features averaged over the K signals. We then design a smooth, non-convex optimization problem to compute a set of signals which are consistent with the estimated averaged features. We find that, in many cases, the proposed approach estimates the set of signals accurately despite non-convexity, and conjecture the number of signals K that can be resolved as a function of the signal length L is on the order of √ L.

Multi-Target Detection With an Arbitrary Spacing Distribution

“…On the other hand, autocorrelation analysis requires shorter computation time by summarizing the data as autocorrelation statistics with one pass over the data. We note that the distinct computational speeds for autocorrelation analysis and the EM algorithm is also observed in the related problem of multi-reference alignment [8], [14]. The difference in run time and the similar reconstruction quality of the two methods at low SNR make autocorrelation analysis the preferred approach for large datasets.…”

Toward single particle reconstruction without particle picking: Breaking the detection limit

“…where W ≥ L is the length of the sought signal and the weights are set to w 1 = 1/2, w 2 = 1/2n 2 , w 3 = 1/2n 3 , where n 2 , n 3 are the number of moments used: n 2 = L − 1, n 3 = (L−1)(L−2) 2 (weights could also be set in accordance with variance estimates as in [9]).…”

Image Recovery from Rotational And Translational Invariants

“…Recall that S 3 ( x 1 , x 2 ) is a discrete version of s 3 ( x 1 , x 2 ), which only depends on the three parameters: | x 1 |, | x 2 | and θ( x 1 , x 2 ). Let S * 3 be the estimate of S 3 from the noisy measurement M via (6), that is, the de-biased and normalized autocorrelation of the measurement A 3 (4), andŜ * 3 denote the DFT of S * 3 . Since the noise is assumed to be i.i.d.…”