Artificial neural network robustness for on-board satellite image processing: results of upset simulations and ground tests

Velazco, Raoul; Cheynet, P.; Muller, Jean-Denis; Ecoffet, R.; Büchner, S.

doi:10.1109/23.659057

Cited by 18 publications

(5 citation statements)

References 11 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The Bayesian-neural-network approach takes full usage of available observations because it obviates the overfitting problem with regularization coefficients, and thus, a subset of data for the cross validation is not necessary (Neal, 1996). In addition, it exhibits robustness to errors in the input parameters (Aires et al, 2004;Velazco et al, 1997), which is important for this study as the input parameters SST and SSS come from gridded data product with errors (section 2.1). The output parameter is ΔfCO 2 .…”

Section: Fco 2 Estimation and Ph Dic And ω Arag Calculationmentioning

confidence: 99%

Long‐Term Changes of Carbonate Chemistry Variables Along the North American East Coast

Cai

Wanninkhof

et al. 2020

JGR Oceans

View full text Add to dashboard Cite

Decadal variability of carbonate chemistry variables has been studied for the open ocean using observations and models, but less is known about the variations in the coastal ocean due to observational gaps and the more complex environments. In this work, we use a Bayesian‐neural‐network approach to reconstruct surface carbonate chemistry variables for the Mid‐Atlantic Bight (MAB) and the South Atlantic Bight (SAB) along the North American East Coast from 1982 to 2015. The reconstructed monthly time series data suggest that the rate of fCO2 increase in the MAB (18 ± 1 μatm per decade) is faster than those in the SAB (14 ± 1 μatm per decade) and the open ocean (14 ± 1 μatm per decade). Correspondingly, pH decreases faster in the MAB. The observed stagnation in the aragonite saturation state, Ωarag decrease during 2005–2015 in the MAB, is attributed to the intrusion of water from southern and offshore regions with high Ωarag, which offsets the decrease expected from anthropogenic CO2 uptake. Furthermore, seasonal asymmetry in the evolution of long‐term change leads to the faster change in the amplitudes of the seasonal cycle in carbonate chemistry variables in coastal waters than those in the open ocean. In particular, the increase in the seasonal‐cycle amplitude of dissolved inorganic carbon in the MAB is 2.9 times larger than that of the open ocean. This leads to the faster increase in the season‐cycle amplitude of Ωarag and earlier occurrence of undersaturation in coastal waters as acidification continues.

show abstract

Section: Fco 2 Estimation and Ph Dic And ω Arag Calculationmentioning

confidence: 99%

Long‐Term Changes of Carbonate Chemistry Variables Along the North American East Coast

Cai

Wanninkhof

et al. 2020

JGR Oceans

View full text Add to dashboard Cite

show abstract

“…The natural redundancy of neural networks and the form of the activation function (usually a sigmoid) of neuron responses make them somewhat fault tolerant, particularly with respect to perturbation patterns. Most of the published work on this topic demonstrated this robustness by injecting limited (Gaussian) noise on a software model [13]. Velazco et al proved the robustness of ANN with respect to bit errors in [13].…”

Section: Artificial Neuralmentioning

confidence: 99%

“…Most of the published work on this topic demonstrated this robustness by injecting limited (Gaussian) noise on a software model [13]. Velazco et al proved the robustness of ANN with respect to bit errors in [13]. Venkitaraman et al proved that neural network architecture exhibits robustness to the input perturbation: the output feature of the neural network exhibits the Lipschitz continuity in terms of the input perturbation in [14].…”

Section: Artificial Neuralmentioning

confidence: 99%

A Robust Data Interpolation Based on a Back Propagation Artificial Neural Network Operator for Incomplete Acquisition in Wireless Sensor Networks

et al. 2018

View full text Add to dashboard Cite

The data space collected by a wireless sensor network (WSN) is the basis of data mining and data visualization. In the process of monitoring physical quantities with large time and space correlations, incomplete acquisition strategy with data interpolation can be adopted to reduce the deployment cost. To improve the performance of data interpolation in such a scenario, we proposed a robust data interpolation based on a back propagation artificial neural network operator. In this paper, a neural network learning operator is proposed based on the strong fault tolerance of artificial neural networks. The learning operator is trained by using the historical data of the data acquisition nodes of WSN and is transferred to estimate the value of physical quantities at the locations where sensors are not deployed. The experimental results show that our proposed method yields smaller interpolation error than the traditional inverse-distance-weighted interpolation (IDWI) method.

show abstract

“…A neural networks based approach was used to classify US images corrupted with speckle noise and estimate the parameters of the distributions of speckle model. Neural networks is also capable of performing a nonlinear processing, constructing a desirable input and output system by learning, and learning a generic model which may be applied to previously unseen examples of input data [19] [20] [21]. Fig.…”

Section: Fuzzy Neural Network With the Epanechnikov Kernelmentioning

confidence: 99%

US image improvement using fuzzy Neural Network with Epanechnikov kernel

Masakuni

Nishimura

2009

2009 35th Annual Conference of IEEE Industrial Electronics

View full text Add to dashboard Cite

Ultrasound (US) imaging is considered as one of the most advanced diagnostic tools in medical use. However, a drawback of the medical US imaging is its poor quality of the image, which is affected by speckle noise. In this paper, a MultiLayer Back-Propagation Neural Networks (MLBPNNs) with the Epanechnikov fuzzy function is proposed to reduce the speckle, and while at the same time, enhance the lesion boundaries of the US image. The main goal of the proposed method is to improve the quality of US image so as to improve the quality of the humans interpretation and the computer systems auto-edge detection. In order to automatically detect the lesion boundary by a computer system, an edge enhancement is required. Evaluating the simulation results by Peak Signal to Noise Ratio (PSNR), Normalized Mean Square Error (NMSE), Detail Variance (DV) ,and Background Variance (BV), the proposed method demonstrates an increased performance of reducing the speckle and enhancing the edge. The proposed method has higher PSNR than conventional methods and can remove the speckle sufficiently, so that tumor boundaries of real US breast tumor image could be preserved and detected.

show abstract

Artificial neural network robustness for on-board satellite image processing: results of upset simulations and ground tests

Cited by 18 publications

References 11 publications

Long‐Term Changes of Carbonate Chemistry Variables Along the North American East Coast

Long‐Term Changes of Carbonate Chemistry Variables Along the North American East Coast

A Robust Data Interpolation Based on a Back Propagation Artificial Neural Network Operator for Incomplete Acquisition in Wireless Sensor Networks

US image improvement using fuzzy Neural Network with Epanechnikov kernel

Contact Info

Product

Resources

About