Fast, Accurate and Memory-Efficient Partial Permutation Synchronization

Li, Shaohan; Shi, Yunpeng

doi:10.1109/cvpr52688.2022.01528

Cited by 5 publications

(3 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With the exception of expensive combinatorial methods [28], [29], existing works consider relaxations to obtain approximate answers. These approximations include spectral relaxation [4], [6], convex relaxation [5], [7], [9], [30], [31], matrix factorization [8], [10], [32], filtering by cluster-consistency statistics [11], message passing [12], and graph clustering [3], [33]- [35]. With the exception of [32], [34], most of the aforementioned methods were originally designed with permutation synchronization in mind and test only with binary associations as input, i.e., in a late fusion approach.…”

Section: A Related Workmentioning

confidence: 99%

“…Consequently, rows of U i are orthogonal, which implies j =r (U i U i ) jr = 0 and therefore U U , P d = 0. Using the standard simplex (12) and the penalty functions ( 14), (15), problem (11) can be equivalently expressed as problem (5), which is repeated below for convenience minimize…”

Section: Appendix I Equivalent Penalty Formmentioning

confidence: 99%

“…Multiway matching attempts to correct outliers by exploiting observation redundancy and enforcing cycle consistency-a property stating that the composition of pairwise matchings over cycles must be identity. Many state-of-theart multiway matching algorithms do this by permutation synchronization [3]- [12] of pairwise correspondences, which are improved and made consistent by joint optimization.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

MIXER: Multiattribute, Multiway Fusion of Uncertain Pairwise Affinities

C.¹,

Fathian²,

How³

2022

Preprint

View full text Add to dashboard Cite

We present a multiway fusion algorithm capable of directly processing uncertain pairwise affinities. In contrast to existing works that require initial pairwise associations, our MIXER algorithm improves accuracy by leveraging the additional information provided by pairwise affinities. Our main contribution is a multiway fusion formulation that is particularly suited to processing non-binary affinities and a novel continuous relaxation whose solutions are guaranteed to be binary, thus avoiding the typical, but potentially problematic, solution binarization steps that may cause infeasibility. A crucial insight of our formulation is that it allows for three modes of association, ranging from non-match, undecided, and match. Exploiting this insight allows fusion to be delayed for some data pairs until more information is available, which is an effective feature for fusion of data with multiple attributes/information sources. We evaluate MIXER on typical synthetic data and benchmark datasets and show increased accuracy against the state of the art in multiway matching, especially in noisy regimes with low observation redundancy. Additionally, we collect RGB data of cars in a parking lot to demonstrate MIXER's ability to fuse data having multiple attributes (color, visual appearance, and bounding box). On this challenging dataset, MIXER achieves 74% F1 accuracy and is 49x faster than the next best algorithm, which has 42% accuracy.

show abstract