Photonic chips with neural networks: Harnessing the light for ultra-fast, low-power artificial Intelligence
The neural networks are distributed computing systems that mimic the brain structure. They aim to provide cognitive performance similar to humans, but in much less time.
These technologies are now the foundation of artificial intelligence and machine learning systems, which can adapt to their environment by analyzing previous actions. They also work autonomously. These technologies are applied in many different areas, including speech and image recognition, augmented reality, autonomous driving, high-performance computing, bioinformatics and genetic and molecular sequences.
In contrast to traditional computing, neural networks require a lot of \”training\” to learn complex functions. The training process is extremely energy intensive and, as computing power increases the neural network’s consumption will grow very quickly, roughly doubling every 6 months.