Chinese scientists have developed a reconfigurable integrated photonic chip capable of running diverse neural networks, including those for image and speech processing, with high efficiency.
Key Details
- 1The chip integrates microring resonators and Mach-Zehnder interferometers, powered by a soliton microcomb light source.
- 2Supports fully-connected, convolutional, and recurrent neural networks within a single hardware architecture.
- 3Area efficiency reaches up to 2.45 TOPS/mm² at 10 GHz frequency.
- 4Demonstrated on tasks: image classification (MNIST 92.93% accuracy, CIFAR-10 56.57%), sentiment analysis (IMDB 80.81%), and speech recognition.
- 5Device enables dual-path computation per resonator, doubling throughput versus traditional schemes.
Why It Matters
This advance shows that photonic computing chips can support the flexible and scalable deployment of multi-model neural networks for imaging and multimodal AI tasks on efficient, high-throughput hardware. Such technology may lead to breakthroughs in hardware acceleration for radiology AI and similar domains.
Source
EurekAlert
Related News
•EurekAlert
AI-Powered OCT Enables Rapid 'Optical Biopsy' for Early Endometrial Cancer Detection
A team at Washington University has developed a catheter-based 3D OCT system with AI to quickly and noninvasively detect early endometrial cancers.
•EurekAlert
AI Clinical Reasoning in Diagnostics and Digital Fatigue in Healthcare
Recent JMIR features explore large language models in clinical diagnostics and digital fatigue among healthcare professionals.
•EurekAlert
KAIST, MIT, Microsoft Develop Efficient AI Image Upsampling for Robotics
KAIST, MIT, and Microsoft have created 'Upsample Anything,' a training-free AI method to restore high-resolution visual data from compressed images with up to 16x improved GPU memory efficiency.