Computing in a flash

• Autores: Timothy Revell
• Localización: New scientist, ISSN 0262-4079, Nº. 3170, 2018, págs. 6-7
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • Artificial intelligence (AI) has a bright future. At last, we have computer processors that use lasers to tackle complex algorithms faster and more efficiently than standard chips can. These optical processors specialize in operations that dominate artificial intelligence, meaning that light could provide the processing power needed to make better, smarter AIs. Optalysys's system has already been used to analyze genomic data, and this week the firm announced that its light processors can run an important type of AI algorithm, called a convolutional neural network. These are currently the go-to algorithm for computer vision tasks, from Facebook's automatic photo tagging features to processing the outside world for autonomous cars. But the algorithms require a lot of computing power and are typically run only on high-performance computers and specialized chips, which are expensive to make and to run.