Search

Two impact-melt fragments in Chang’e-6 lunar soils reveal that the Apollo basin was formed at ~4.16 Ga, implying a non-cataclysmic late heavy bombardment on the Moon between 3.8 and 4.0 Ga.

The spatiotemporal relationship between neuromodulator and neurotransmitter dynamics and neuronal activity remains unclear. Here, the authors performed dual-color imaging of calcium and ACh/5-HT signals across the fly brain, revealing a key role of ACh dynamics in stable odor representation.

This study reports a complete photonic neuron integrated on a silicon-nitride chip, enabling ultrafast all-optical computing with nonlinear multi-kernel convolution for image recognition and motion generation.

A photonic processor based on a diffractive tensorized unit enables million-TOPS general-purpose computing. The approach challenges the generality and scalability constraints of diffractive computing and enables orders-of-magnitude improvements in energy efficiency over a high-end electronic tensor core processor.

The integration of photonic components on silicon chips creates the challenge of achieving a uniform and efficient architecture. Here, the authors demonstrate on-chip light-emitters, photodetectors, photovoltaic power supply and optical data link, all based on InGaAs nanoresonators grown on silicon.

We present fully forward mode learning, which conducts machine learning operations on site, leading to faster learning and promoting advancement in numerous fields.

A wide-field wavefront sensor consisting of a microlens array on the native image plane enables observation of atmospheric turbulence over a field of view of 1,100 arcsec at 30 Hz with an 80 cm telescope. With the aid of a neural network, turbulence can be predicted 33 ms in advance.

DeepWonder removes background signals from widefield calcium recordings and enables accurate and efficient neuronal segmentation with high throughput.

A meta-imaging sensor detects an extra-fine 4D light field distribution using a vibrating microlens array, enabling high-resolution 3D photography up to a gigapixel with fast aberration correction, demonstrated on a telescope aimed at the Moon.

Optoelectronic neural networks are a promising avenue in AI computing for parallelization, power efficiency, and speed. Here, the authors present a dual-neuron optical-artificial learning approach for training large-scale diffractive neural networks, achieving VGG-level performance on ImageNet in simulation with a network that is 10 times larger than existing ones.

An all-analog chip combining electronic and light computing achieves systemic energy efficiency of more than three orders of magnitude and a computing speed of more than one order of magnitude compared with state-of-the-art computing processors.

DeepCAD is a self-supervised deep-learning approach for denoising calcium imaging data. DeepCAD improved SNR and facilitates neuron extraction and spike inference.

Traditional optical microscope, while bulky, often fails to deliver optimal performance. Here, the authors have engineered an integrated microscope of 0.15 cm³ in volume and a weight of 0.5 g, which outperforms a commercial microscope and can be seamlessly integrated with a smartphone.

Vascular abnormalities are challenging for diagnostic imaging due to the complexity of vasculature and the non-uniform scattering from biological tissues. The authors present an unsupervised learning algorithm for vascular feature recognition from small sets of biomedical images acquired from different modalities. They demonstrate the utility of their diagnostic approach on vascular images of thrombosis, internal bleeding and colitis.

DeepCAD-RT denoises fluorescence time-lapse images in real time.

Linear diffractive structures are by themselves passive systems but researchers here exploit the non-linearity of a photodetector to realize a reconfigurable diffractive ‘processing’ unit. High-speed image and video recognition is demonstrated.

Mirror-enhanced scanning light-field microscopy (MiSLFM) achieves long-term high-speed 3D imaging at near-isotropic subcellular resolution with a single objective, facilitating native observations of various intracellular and intercellular interactions.

Inspired by human brain for multi-task continual learning, a generalized photonic neuromorphic architecture (L²ONN) is proposed to model physical-driven light sparsity and parallelism, towards reconfigurable and scalable lifelong learning.

We develop Monet: a multichannel optical neural network architecture for a universal multiple-input multiple-channel optical computing based on a novel projection-interference-prediction framework, solving real-world advanced machine vision tasks optically.