Encoding a priori information in neural networks to improve its modeling performance under non-stationary environment

This paper proposes a new methodology to approximate functions by incorporating a priori information. The relationship between the proposed scheme and multilayer neural networks is explored theoretically and numerically. This approach is particularly interesting for the very relevant class of limited spectrum functions. The number of free parameters is smaller if compared to Back-Propagation Algorithm opening the way for better generalization results.

Download Full-text

Estimation of soil parameters over bare agriculture areas from C-band polarimetric SAR data using neural networks

Hydrology and Earth System Sciences ◽

10.5194/hess-16-1607-2012 ◽

2012 ◽

Vol 16 (6) ◽

pp. 1607-1621 ◽

Cited By ~ 57

Author(s):

N. Baghdadi ◽

R. Cresson ◽

M. El Hajj ◽

R. Ludwig ◽

I. La Jeunesse

Keyword(s):

Neural Networks ◽

Surface Roughness ◽

Soil Moisture ◽

A Priori ◽

Soil Parameters ◽

A Priori Information ◽

Polarimetric Sar ◽

The Neural Networks ◽

Sar Data ◽

Priori Information

Abstract. The purpose of this study was to develop an approach to estimate soil surface parameters from C-band polarimetric SAR data in the case of bare agricultural soils. An inversion technique based on multi-layer perceptron (MLP) neural networks was introduced. The neural networks were trained and validated on a noisy simulated dataset generated from the Integral Equation Model (IEM) on a wide range of surface roughness and soil moisture, as it is encountered in agricultural contexts for bare soils. The performances of neural networks in retrieving soil moisture and surface roughness were tested for several inversion cases using or not using a-priori knowledge on soil parameters. The inversion approach was then validated using RADARSAT-2 images in polarimetric mode. The introduction of expert knowledge on the soil moisture (dry to wet soils or very wet soils) improves the soil moisture estimates, whereas the precision on the surface roughness estimation remains unchanged. Moreover, the use of polarimetric parameters α1 and anisotropy were used to improve the soil parameters estimates. These parameters provide to neural networks the probable ranges of soil moisture (lower or higher than 0.30 cm3 cm−3) and surface roughness (root mean square surface height lower or higher than 1.0 cm). Soil moisture can be retrieved correctly from C-band SAR data by using the neural networks technique. Soil moisture errors were estimated at about 0.098 cm3 cm−3 without a-priori information on soil parameters and 0.065 cm3 cm−3 (RMSE) applying a-priori information on the soil moisture. The retrieval of surface roughness is possible only for low and medium values (lower than 2 cm). Results show that the precision on the soil roughness estimates was about 0.7 cm. For surface roughness lower than 2 cm, the precision on the soil roughness is better with an RMSE about 0.5 cm. The use of polarimetric parameters improves only slightly the soil parameters estimates.

Download Full-text

Soil parameters estimation over bare agriculture areas from C-band polarimetric SAR data using neural networks

Hydrology and Earth System Sciences Discussions ◽

10.5194/hessd-9-2897-2012 ◽

2012 ◽

Vol 9 (3) ◽

pp. 2897-2933 ◽

Cited By ~ 2

Author(s):

N. Baghdadi ◽

R. Cresson ◽

M. El Hajj ◽

R. Ludwig ◽

I. La Jeunesse

Keyword(s):

Neural Networks ◽

Surface Roughness ◽

Soil Moisture ◽

A Priori ◽

Soil Parameters ◽

A Priori Information ◽

Polarimetric Sar ◽

The Neural Networks ◽

Sar Data ◽

Priori Information

Abstract. The purpose of this study was to develop an approach to estimate soil surface parameters from C-band polarimetric SAR data in the case of bare agricultural soils. An inversion technique based on Multi-Layer Perceptron (MLP) neural networks was introduced. The neural networks were trained and validated on a noisy simulated dataset generated from the Integral Equation Model (IEM) on a wide range of surface roughness and soil moisture, as it is encountered in agricultural contexts for bare soils. The performances of neural networks in retrieving soil moisture and surface roughness were tested for several inversion cases in using or not a priori knowledge on soil parameters. The inversion approach was then validated in using RADARSAT-2 images in polarimetric mode. The introduction of expert knowledge on the soil moisture (dry to wet soils or very wet soils) improves the soil moisture estimates whereas the precision on the surface roughness estimation remains unchanged. Moreover, the use of polarimetric parameters α1 and anisotropy were used to improve the soil parameters estimates. These parameters provide to neural networks the probable ranges of soil moisture (lower or higher than 0.30 cm3 cm−3) and surface roughness (lower or higher than 1.5 cm). Soil moisture can be retrieved correctly from C-band SAR data by using the neural networks technique. Soil moisture errors were estimated at about 0.098 without a priori information on soil parameters and 0.065 cm3 cm−3 (RMSE) applying a priori information on the soil moisture. The retrieval of surface roughness is possible only for low and medium values (lower than 2 cm). Results show that the precision on the soil roughness estimates was about 0.7 cm. For surface roughness lower than 2 cm, the precision on the soil roughness is better with a RMSE about 0.5 cm. The use of polarimetric parameters improves only slightly the soil parameters estimates.

Download Full-text

Planning of technical objects and systems

International Scientific and Technical conference "The Progressive Technics, Technology and Engineering Education" ◽

10.20535/2409-7160.2021.xxii.240483 ◽

2021 ◽

Author(s):

R. Zinko

Keyword(s):

Neural Network ◽

Neural Networks ◽

Process Management ◽

A Priori ◽

Self Organization ◽

A Priori Information ◽

Priori Information ◽

Physical Phenomena ◽

Structural Commonality ◽

Reference Functions

There are many types and methods of simulation, but among them special attention should be paid to methods based on the theory of heuristic self - organization. All algorithms of the method of group argumentation (MGVA) are characterized by structural commonality on the principle of self - organization, which require insignificant requirements for a priori information to search for an infinite number of options. The advantage of the algorithm of the method of group consideration of arguments in comparison with other algorithms of this class is the presence of possibilities of expansion of the vector of initial data and the device for elimination of collinearity - reception of orthogonalization. MGVA consists of two blocks: pre - processing of observations taking into account the system of selected reference functions and calculation of selection applicants. As a result of the algorithm, models capable of controlling the process taking into account the phenomena accompanying a certain process are obtained. Given the commonality of the main provisions of the theory of self - organization of artificial neural networks and MGVA, the network variables are added to the model as a variable Z. As a result, we obtain a neural network that describes the physical phenomena accompanying the process. This will significantly increase the efficiency and accuracy of process management.

Download Full-text