Smoothed-NUV Priors for Imaging

Ma, Boxiao; Zalmai, Nour; Loeliger, Hans-Andrea

doi:10.1109/tip.2022.3186749

Cited by 1 publication

(3 citation statements)

References 76 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…With that generalization in mind, we now consider computing (48) by AM over x and θ. As in Section III-A1, we precompute the maximization over θ by plugging the maximizing variances ( 44) into (18), resulting in the closedform update rules…”

Section: Binarizing Priormentioning

confidence: 99%

“…The derivation of (62) and ( 63) is detailed in Appendix C. Again, the maximizing variances (62) and (63) can be plugged into (18), resulting in where we have dropped the iteration index for readibility. The update rules (64) and ( 65) are given in Table II (in slightly different notation).…”

Section: Binarizing Priormentioning

confidence: 99%

“…The point of such priors is the computational compatibility with linear Gaussian models. The primary use of such priors has been to encourage sparsity, in applications including sparse input estimation [8], [9], localized event detection [10], [11], [12], outlier removal [13], [14], sparse least squares [7], control [15], [16], and imaging [17], [18].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Binary Control and Digital-to-Analog Conversion Using Composite NUV Priors and Iterative Gaussian Message Passing

Keusch

Malmberg

Loeliger

2021

ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

The paper proposes a new method to determine a binary control signal for an analog linear system such that the state, or some output, of the system follows a given target trajectory. The method can also be used for digital-to-analog conversion.The heart of the proposed method is a new binaryenforcing NUV prior (normal with unknown variance). The resulting computations, for each planning period, amount to iterating forward-backward Gaussian message passing recursions (similar to Kalman smoothing), with a complexity (per iteration) that is linear in the planning horizon. In consequence, the proposed method is not limited to a short planning horizon.Index Terms-Discrete-level priors, normals with unknown variance (NUV), finite-control-set model predictive control (MPC), digital-to-analog conversion (DAC).

show abstract

Section: Binarizing Priormentioning

confidence: 99%

Section: Binarizing Priormentioning

confidence: 99%