Issue link: https://fhpublishing.uberflip.com/i/808182
NETWORK / 29 / APRIL 2017 The project received £460,000 funding from the Network Innovation Allowance. 5. Cognitive learning It's not just the technologies currently being trialled by networks that have the potential to revolutionise the energy grid. Computers used to be regarded as machines that could "think" extremely fast but lacked intuition and creativeness. That changed in 1997 when the IBM-built computer Deep Blue II beat the then world chess champion Gary Kasparov at his own game. Deep Blue II could evaluate 200 million positions per second, and typically search to a depth of between six and eight moves to a maximum of 20 or even more in some situations. This kind of depth of "thought" and com- puting power typi‹ es modern AI and cogni- tive technology. DeepMind, for example, is working on a machine that can retain knowledge and "learn" incrementally, rather than "cata- strophic forgetting", which typi‹ es current deep neural networks. to customers. N When IBM's Deep Blue II beat world champion Gary Kasparov in a chess tournament, it raised AI to another level. This means an AI would be able to see patterns of behaviours and devise ways to minimise energy consumption in large-scale commercial and industrial systems such as data centres. For example, DeepMind's machine learning algorithms have already helped Google cut energy use by 40% by being able to predict the temperature and pressure outputs within its data cen- tres 60 minutes in advance. Roll this out to all UK data centres, for example, and the impact of electric vehicle charging on demand could be mitigated, deferring rein- forcement costs and ensur- ing the best possible service For example, DeepMind's machine learning algorithms have already helped Google cut energy use by 40% by being the impact of electric vehicle charging on demand could ing the best possible service