On the Importance of α Marginalization in Maximum Entropy
- 637 Downloads
The correct entropic prior, computed by marginalization over the reg- ularization parameter a, is used to invert photoemission data and to restore the famous “Susie” image. Comparison with the conventional maximum entropy procedure shows less overfitting of noise and demonstrates the residual ringing which is intrinsic to ill-posed inversion problems. An improvement to the steepest descent approximation reveals the reason for the overfitting. On top of that, the correct treatment of the regularization parameter is vital for the existence of the continuum-limit of MaxEnt.
Key wordsInverse Problem Regularization Entropic Prior Image Processing
Unable to display preview. Download preview PDF.
- 1.S. F. Gull and J. Skilling, “Maximum entropy method in image processing,” IEE Proc, 131 F, pp. 646–659, 1984.Google Scholar
- 2.C. E. M. Strauss, D. H. Wolpert and D. R. Wolf, “Alpha, evidence, and the entropic prior,” in Maximum Entropy and Bayesian Methods (1993), G. Heidbreder, ed., Kluwer Academic, Dordrecht, 1995.Google Scholar
- 3.S. F. Gull, “Developments in maximum entropy data analysis,” in Maximum Entropy and Bayesian Methods, J. Skilling, ed., pp. 53–71, Kluwer Academic, Dordrecht, 1989.Google Scholar
- 6.W. von der Linden, R. Fischer, and V. Dose, “Evidence integrals,” In these proceedings, 1995.Google Scholar
- 7.J. Skilling and S. Sibisi, “Prior distributions on measure space,” In these proceedings, 1995.Google Scholar
- 8.J. Skilling, “Fundamentals of MaxEnt in data analysis,” in Maximum Entropy in Action, B. Buck and V. A. Macaulay, eds., pp. 19–40, Clarendon Press, Oxford, 1991.Google Scholar
- 9.R. Puetter, “Pixon-based image reconstruction in high energy astrophysics,” In these proceedings, 1995.Google Scholar
- 12.R. Fischer, “submitted to Surf. Sci.”Google Scholar