AI Training has an Energy Problem and Nanomagnetic Computing is Here to Help

Researchers have invented a new method via nanomagnets in which AI training will not be energy-intensive

Training advanced AI and Ml models takes immense time and money to ensure that the data is of high quality, but AI training requires a significant amount of energy that has an adverse effect on the environment. Between storing data in large data centers, then using it to train an ML program, the process requires exceptional amounts of energy and is highly condemned by environmentalists. But recently, researchers have found a new method to minimize the energy cost of training an AI model.

A team of researchers led by Imperial College London invented a new method to slash the current globally doubling energy consumption. In a paper published today in Nature Nanotechnology, the international team has claimed that the first proof is that networks of nanomagnets can be used to perform AI-like processing. The researchers showed nanomagnets can be used for 'time-series prediction tasks, such as predicting and regulating insulin levels in diabetic patients.

Coming to the role of nanomagnets in this, experts said that professionals can now train AI to do even relatively simple tasks without spending huge amounts of energy. Much of the energy used to achieve this would traditionally be wasted in insufficient transport of electrons during processing and memory storage. But nanomagnets do not rely on the physical transport of particles like electrons, but instead process and transfer information in the form of a 'magnon' wave, where every magnet is affected by the state of the neighbor magnets.

For the next part of the research, the team has decided to teach the system using real-world data, such as ECG signals, and hopes to make it into a real computing device. Magnetic systems could then be eventually integrated into conventional computers to improve energy efficiency for intense processing tasks.