In order to participate in the seminar, on June 11 after 4:45 pm (UTC+7) you should connect to the Zoom conference via the following link or manually using the Zoom conference ID 897 7646 2466 and password 549526.One of the most popular topics at the intersection of data analysis and optimization lately is how to train deep neural networks. Mathematically, the problem is reduced to the problem of stochastic optimization, which, in turn, using the Monte Carlo method, is reduced to the problem of minimizing the sum of a large number of functions. It is important to note that a similar plot is inherent in general for almost all tasks that come from data analysis. Almost all data analysis (machine learning) problems are reduced to optimization problems, or rather stochastic optimization. In mathematical statistics with a known probability law (but unknown parameters), and in machine learning 鈥� with an unknown probability law. One of the most popular ways to solve such optimization problems and their variants obtained using the Monte Carlo method is the stochastic gradient descent method and its variations. The methods were known back in the 1950s. However, the real value of this method has been appreciated in the last twenty years in connection with the noted applications. In this talk I will make a small overview of the development of this direction in recent years (adaptive selection of a step, batch size, federated learning, etc.).
