Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain
the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in
Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles
and JavaScript.
Using 760,000 phase-change memory devices, an in-memory hyperdimensional computing system can be created that can perform machine learning tasks such language classification. Hyperdimensional computing takes inspiration from the brain and uses the mathematical properties of hyperdimensional spaces comprised of high-dimensional binary vectors known as hypervectors. The schematic illustration on the cover highlights these hypervectors (represented as cylinders) and the quasi-orthogonality of large hypervectors; the elements of the hypervectors are encoded using different atomic arrangements of small volumes of phase-change material (represented here with different ball-and-stick structures).
The adoption of autonomous vehicles will depend on the public’s trust in the technology. To establish such trust, it is important to understand the actions of different road users, and their reactions to the vehicles.
Modern cars have an array of sensors that allow different objects to be recognized, including large and small animals. They thus have the potential to become a tool for monitoring biodiversity and improving driver safety. But to achieve this various challenges in computing, communications and privacy need to be addressed.
A fibre with liquid-metal core and soft outer shell can be woven into textiles and used to sense multiple compression and stretching events simultaneously.
A magnetoresistance effect that occurs in a platinum layer deposited on a magnon junction consisting of two insulating magnetic yttrium iron garnet layers separated by an antiferromagnetic nickel oxide spacer layer could be used to create spintronic and magnonic devices that are free from Joule heating.
A favourable scaling law for the magnetic state readout of CoFe/Pt nanostructure devices allows large spin Hall signals of 0.3 Ω at room temperature to be obtained, which could be useful in the development of spin-logic devices.
A microstructured liquid metal fibre can be used as a soft transmission line probe for reflectometry, allowing electronic textiles to be created that can decipher convoluted mechanical stimulation.
A complete in-memory hyperdimensional computing system, which uses 760,000 phase-change memory devices, can efficiently perform machine learning related tasks including language classification, news classification and hand gesture recognition from electromyography signals.
The monolithic integration of electronic and plasmonic technologies can be used to create electro-optic transmitters capable of symbol rates beyond 100 GBd.