TY - JOUR
T1 - Unrolled Algorithms for Group Synchronization
AU - Janco, Noam
AU - Bendory, Tamir
N1 - Publisher Copyright:
Author
PY - 2023
Y1 - 2023
N2 - The group synchronization problem involves esti-mating a collection of group elements from noisy measurements of their pairwise ratios. This task is a key component in many computational problems, including the molecular reconstruction problem in single-particle cryo-electron microscopy (cryo-EM). The standard methods to estimate the group elements are based on iteratively applying linear and non-linear operators, and are not necessarily optimal. Motivated by the structural similarity to deep neural networks, we adopt the concept of algorithm unrolling, where training data is used to optimize the algorithm. We design unrolled algorithms for several group synchronization instances, including synchronization over the group of 3-D rota-tions: the synchronization problem in cryo-EM. We also apply a similar approach to the multi-reference alignment problem. We show by numerical experiments that the unrolling strategy outperforms existing synchronization algorithms in a wide variety of scenarios.
AB - The group synchronization problem involves esti-mating a collection of group elements from noisy measurements of their pairwise ratios. This task is a key component in many computational problems, including the molecular reconstruction problem in single-particle cryo-electron microscopy (cryo-EM). The standard methods to estimate the group elements are based on iteratively applying linear and non-linear operators, and are not necessarily optimal. Motivated by the structural similarity to deep neural networks, we adopt the concept of algorithm unrolling, where training data is used to optimize the algorithm. We design unrolled algorithms for several group synchronization instances, including synchronization over the group of 3-D rota-tions: the synchronization problem in cryo-EM. We also apply a similar approach to the multi-reference alignment problem. We show by numerical experiments that the unrolling strategy outperforms existing synchronization algorithms in a wide variety of scenarios.
KW - algorithm unrolling
KW - Group synchronization
KW - Maximum likelihood estimation
KW - multi-reference alignment
KW - Noise measurement
KW - Signal processing algorithms
KW - Signal to noise ratio
KW - Standards
KW - Synchronization
KW - Training data
UR - http://www.scopus.com/inward/record.url?scp=85170546733&partnerID=8YFLogxK
U2 - 10.1109/OJSP.2023.3311354
DO - 10.1109/OJSP.2023.3311354
M3 - ???researchoutput.researchoutputtypes.contributiontojournal.article???
AN - SCOPUS:85170546733
SN - 2644-1322
SP - 1
EP - 12
JO - IEEE Open Journal of Signal Processing
JF - IEEE Open Journal of Signal Processing
ER -