Machine Learning Seminar by Andrew Mundy, Arm Research

Today, Dr Andrew Mundy from Arm will be giving a talk at the University on the activities of the machine learning group at Arm Research.

Title: Machine Learning at ARM Research (An introduction to the Bonseyes project and Arm Research)

Abstract: Bonseyes, a Horizon 2020 funded project, is a response to the growth of end-to-end solutions for systems of Artificial Intelligence and Machine Learning. These end-to-end solutions favour large internet companies, such as Google or Facebook, who have come to rapidly dominate the AI marketplace. Bonseyes attempts to break this dominance by challenging the vicious cycle that developing effective systems of artificial intelligence requires vast amounts of data but that acquiring vast amounts of data typically requires a large user base, which itself requires the existence of a good product. Specifically, Bonseyes attempts to foster collaboration between different actors to allow sharing of data and model building expertise. Several outcomes are planned: a data and model marketplace which enables sharing and collaboration in the production of AI systems; a toolbox of edge-device-aware training and network design methods which will allow the rapid creation of energy-efficient and low-cost neural networks; and the construction of developer reference platforms to reduce the cost of entry to building new AI systems. In this talk I will present an overview of the project and highlight several interesting results achieved to date. Additionally, I will give an overview of the work of Arm Research.

Speaker: Andrew Mundy is a member of the Machine Learning group at Arm Research. After gaining a Bachelor’s degree in Electrical and Electronic Engineering at Newcastle University in 2009 he joined the SpiNNaker group at the University of Manchester to complete his PhD on real-time cognitive modelling of the brain using the Neural Engineering Framework.