Mathematics Colloquia and Seminars

Return to Colloquia & Seminar listing

In a variety of fields such as signal processing and data mining, recent developments (such as high resolution imaging or big data generation) have produced problems much larger than can be efficiently solved by gradient descent or interior point methods. Thus we seek quasi-Newton methods, which balance gradient descent efficiency with Newton method accuracy. We will discuss current quasi-Newton methods which build on each other intuitively and arrive at the current fastest method for solving a class of QPs, one which has interestingly been derived by two different authors in very different ways.