| Article ID: | iaor1990279 |
| Country: | United States |
| Volume: | 14 |
| Issue: | 4 |
| Start Page Number: | 745 |
| End Page Number: | 759 |
| Publication Date: | Nov 1989 |
| Journal: | Mathematics of Operations Research |
| Authors: | Birge John, Teboulle Marc |
| Keywords: | gradient methods |
New upper bounds are given for the expected value of a convex function. The bounds employ subgradient information and the conjugate function. In contrast to most other bounds, explicit moment information is not needed. The authors derive the bounds and compare them with previous bounds with different information requirements.