Regression is a well-known statistical modeling technique to establish a meaningful relationship between independent and dependent variables.

The development of machine-learning algorithms with real or continuous -valued output data gives rise to regression problems, which in turn, require techniques of optimizations for their solutions.

After giving a brief overview of the subject of machine-learning, it will be shown how and what sort of regression problems arise in the study of machine-learning,   and then discuss in some details numerically viable algorithms for optimization which can be gainfully employed to effectively solve these regressions problems.

The talk is rather elementary in nature and interdisciplinary. It should be accessible and beneficial to a wide variety of audience, ranging from mathematics, computer science, statistics, medical and health sciences, signal and image processing engineering, and many others. In particular, it should provide a strong incentive to the educators to develop an interdisciplinary course or curriculum along the line of the subject matter of this talk.