תקציר
The Misspecified Cramér-Rao lower bound (MCRB) provides a lower bound on the performance of any unbiased estimator of parameter vector θ under model misspecification. An approximation of the MCRB can be numerically evaluated using a set of i.i.d samples of the true distribution at θ. However, obtaining a good approximation for multiple values of θ requires collocating an unrealistically large number of samples. In this paper, we present a method for approximating the MCRB using a Generative Model, referred to as a Generative Misspecified Lower Bound (GMLB), in which we train a generative model on data from the true measurement distribution. Then, the generative model can generate as many samples as required for any θ, and therefore the GMLB can use a limited set of training data to achieve an excellent approximation of the MCRB for any parameter. We demonstrate the GMLB on two examples: a misspecified Linear Gaussian model; and a Non-Linear Truncated Gaussian model. In both cases, we empirically show the benefits of the GMLB in accuracy and sample complexity. In addition, we show the ability of the GMLB to approximate the MCRB on unseen parameters.
שפה מקורית | אנגלית |
---|---|
כותר פרסום המארח | ICASSP 2023 - 2023 IEEE International Conference on Acoustics, Speech and Signal Processing, Proceedings |
מוציא לאור | Institute of Electrical and Electronics Engineers Inc. |
עמודים | 1-5 |
מספר עמודים | 5 |
מסת"ב (אלקטרוני) | 9781728163277 |
מזהי עצם דיגיטלי (DOIs) | |
סטטוס פרסום | פורסם - 2023 |
פורסם באופן חיצוני | כן |
אירוע | 48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023 - Rhodes Island, יוון משך הזמן: 4 יוני 2023 → 10 יוני 2023 |
סדרות פרסומים
שם | ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings |
---|---|
כרך | 2023-June |
ISSN (מודפס) | 1520-6149 |
כנס
כנס | 48th IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2023 |
---|---|
מדינה/אזור | יוון |
עיר | Rhodes Island |
תקופה | 4/06/23 → 10/06/23 |
הערה ביבליוגרפית
Publisher Copyright:© 2023 IEEE.