Philippe Massicotte | Searcharxiv

arXiv2026

Inversion of Sea Ice Spectral Albedo to Estimate Under-Ice Transmittance

Sunlight radiation under snow-covered sea ice obtained from remote sensing could help assess under-ice primary production at pan-Arctic scale. Yet, the current remote sensing methods to estimate sunlight transmittance under sea ice is limited by its reliance on imprecise snow depth products and its inability to sense microstructure-driven variations in snow and ice light scattering properties. Based on Monte-Carlo simulations of radiative transfer, we developed an inversion method relying solely on spectral albedo to estimate transmittance under snow-covered sea ice. The method analyses albedo spectral information to derive the vertically resolved scattering properties of snow and sea ice above the freeboard. Assuming fixed columnar ice physical and optical properties, transmittance is then estimated. At ground level, our spectral albedo inversion method is more precise than the current approaches. We argue this is because it implicitly accounts for the variability in snow scattering properties. This method could significantly improve the satellite estimation of photosynthetically available radiation under sea ice, especially because it does not need snow depth.

physics.ao-ph↗

arXiv2019

S-ConvNet: A Shallow Convolutional Neural Network Architecture for Neuromuscular Activity Recognition Using Instantaneous High-Density Surface EMG Images

The concept of neuromuscular activity recognition using instantaneous high-density surface electromyography (HD-sEMG) images opens up new avenues for the development of more fluid and natural muscle-computer interfaces. However, the existing approaches employed a very large deep convolutional neural network (ConvNet) architecture and complex training schemes for HD-sEMG image recognition, which requires the network architecture to be pre-trained on a very large-scale labeled training dataset, as a result, it makes computationally very expensive. To overcome this problem, we propose S-ConvNet and All-ConvNet models, a simple yet efficient framework for learning instantaneous HD-sEMG images from scratch for neuromuscular activity recognition. Without using any pre-trained models, our proposed S-ConvNet and All-ConvNet demonstrate very competitive recognition accuracy to the more complex state of the art for neuromuscular activity recognition based on instantaneous HD-sEMG images, while using a ~ 12 x smaller dataset and reducing learning parameters to a large extent. The experimental results proved that the S-ConvNet and All-ConvNet are highly effective for learning discriminative features for instantaneous HD-sEMG image recognition especially in the data and high-end resource constrained scenarios.

eess.SP↗