In this thesis we study the convergence rates and complexities of a continuous model of the Stochastic Gradient Descent (SGD) under convexity, strong convexity and Łojasiewicz assumptions, the latter being a way to generalize ...