The number of measurement operator evaluations used by the quantifAI for the W28 image. We do not distinguish between the measurement operator and its adjoint. Therefore, evaluating the log-likelihood gradient counts as two evaluations of the measurement operator. The fast pixel UQ is three and six orders of magnitude faster than the MCMC sampling and LCIs, respectively.
MCMC . | LCIs . | LCIs . | Fast . |
---|---|---|---|
sampling . | |$8 \times 8$| . | |$16 \times 16$| . | pixel UQ . |
|$11 \times 10^{6}$| | |$81.5 \times 10^{3}$| | |$21.2 \times 10^{3}$| | 28 |
MCMC . | LCIs . | LCIs . | Fast . |
---|---|---|---|
sampling . | |$8 \times 8$| . | |$16 \times 16$| . | pixel UQ . |
|$11 \times 10^{6}$| | |$81.5 \times 10^{3}$| | |$21.2 \times 10^{3}$| | 28 |
The number of measurement operator evaluations used by the quantifAI for the W28 image. We do not distinguish between the measurement operator and its adjoint. Therefore, evaluating the log-likelihood gradient counts as two evaluations of the measurement operator. The fast pixel UQ is three and six orders of magnitude faster than the MCMC sampling and LCIs, respectively.
MCMC . | LCIs . | LCIs . | Fast . |
---|---|---|---|
sampling . | |$8 \times 8$| . | |$16 \times 16$| . | pixel UQ . |
|$11 \times 10^{6}$| | |$81.5 \times 10^{3}$| | |$21.2 \times 10^{3}$| | 28 |
MCMC . | LCIs . | LCIs . | Fast . |
---|---|---|---|
sampling . | |$8 \times 8$| . | |$16 \times 16$| . | pixel UQ . |
|$11 \times 10^{6}$| | |$81.5 \times 10^{3}$| | |$21.2 \times 10^{3}$| | 28 |
This PDF is available to Subscribers Only
View Article Abstract & Purchase OptionsFor full access to this pdf, sign in to an existing account, or purchase an annual subscription.