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
MD Anderson Unveils New AI Genomics Insights and Therapeutic Advances
MD Anderson reports breakthroughs in cancer therapeutics and provides critical insights into AI models for genomic analysis.
•EurekAlert
UCLA Researchers Present AI, Blood Biomarker Advances at SABCS 2025
UCLA Health researchers unveil major advances in breast cancer AI pathology, liquid biopsy, and biomarker strategies at the 2025 SABCS.
•EurekAlert
SH17 Dataset Boosts AI Detection of PPE for Worker Safety
University of Windsor researchers released SH17, a 8,099-image open dataset for AI-driven detection of personal protective equipment (PPE) in manufacturing settings.