Events

Paulo Oliva: Stochastic Approximation. Fundamentals of AI reading group

Centre for Fundamentals of AI and Computational Theory 
Image: An illustration of Stochastic Approximation. The equation of the curve M(x) is unknown other than by sampling which gives approximate values, but we want to find the point θ where the curve has some given value α.
An illustration of Stochastic Approximation. The equation of the curve M(x) is unknown other than by sampling which gives approximate values, but we want to find the point θ where the curve has some given value α.

Date: 18 March 2026   Time: 10:30 - 12:00

Location: Engineering G2, Mile End, E1 4NS Map 

Paulo Oliva will talk about Stochastic Approximation, from its birth in 1952 (Robbins-Monro method), through Dvoretzky's generalisation (1956), the development of stochastic gradient descent (Kiefer-Wolfowitz and Rosenblatt, 1958), and the formalisation of backpropagation (1986), up to more recent developments.

After Paulo's talk and discussions, there will be lunch from 12:00.

Contact:  Frederik Dahlqvist
Email:  f.dahlqvist@qmul.ac.uk

Updated by: Paul Curzon