Neural network based solution to inverse problems

Ogawa, Takehiko; Kosugi, Yukio; Kanada, Hajime

doi:10.1109/ijcnn.1998.687250

Cited by 10 publications

(4 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…These techniques all require training of the neural network in the forward direction and are dual port networks (input and output). In contrast, Ogawa et al [ 10 ] have suggested a triple port network with weights being the third one. The inverse solution is thus extracted from this third port.…”

Section: Introductionmentioning

confidence: 99%

Deep neural-network based optimization for the design of a multi-element surface magnet for MRI applications

2022

View full text Add to dashboard Cite

We present a combination of a CNN-based encoder with an analytical forward map for solving inverse problems. We call it an encoder-analytic (EA) hybrid model. It does not require a dedicated training dataset and can train itself from the connected forward map in a direct learning fashion. A separate regularization term is not required either, since the forward map also acts as a regularizer. As it is not a generalization model it does not suffer from overfitting. We further show that the model can be customized to either finding a specific target solution or one that follows a given heuristic. As an example, we apply this approach to the design of a multi-element surface magnet for low-field magnetic resonance imaging (MRI). We further show that the EA model can outperform the benchmark genetic algorithm model currently used for magnet design in MRI, obtaining almost 10 times better results.

show abstract

Section: Introductionmentioning

confidence: 99%

Deep neural-network based optimization for the design of a multi-element surface magnet for MRI applications

2022

View full text Add to dashboard Cite

show abstract

“…As opposed to direct problem, the wellposedness of the inverse problem, i.e. the uniqueness of the solution, is not always guaranteed (Ogawa et al (1998)). …”

Section: Introductionmentioning

confidence: 99%

A novel inverse dynamic model for a magnetorheological damper based on network inversion

Boada

Díaz

2017

Journal of Vibration and Control

View full text Add to dashboard Cite

Semi-active suspensions based on magnetorheological (MR) dampers are receiving significant attention specially for control of vibration isolation systems. The nonlinear hysteretic behaviour of MR dampers can cause serious problems in controlled systems such as instability and loss of robustness. Most of the developed controllers determine the desired damping forces which should be produced by the MR damper. Nevertheless, the MR damper behaviour can only be controlled in terms of the applied current (or voltage). In addition to this, it is necessary to develop an adequate inverse dynamic model in order to calculate the command current (or voltage) for the MR damper to generate the desired forces as close as possible to the optimal ones. Due to MR dampers are highly nonlinear devices, the inverse dynamics model is difficult to obtain. In this paper, a novel inverse MR damper model based on a network inversion to estimate the necessary current (or voltage) such as the desired force is exerted by the MR damper is presented. The proposed inverse model is validated carrying out experimental tests. In addition, a comparison of simulated tests with other damper controllers is also presented. Results show the effectiveness of the network inversion for inverse modeling of an MR damper, so that the proposed inverse model can act as a damper controller to generate the command current (or voltage) to track the desired damping force. Keywords

show abstract

“…We can also use regularization for network inversion [17]. In order to provide the constraint condition, we minimize the regularization functional in accordance with the output error in the inverse estimation phase.…”

Section: Ill-posedness and Regularizationmentioning

confidence: 99%

“…The regularization method to decrease illposedness by limiting the solution space of the inverse problem has been proposed for the ill-posed inverse problems [14,15]. It has also been examined for network inversion [16,17]. The inverse-modeling approach with a multilayer neural network and the approach that uses the network inversion method have been examined as neural-networkbased methods.…”

Section: Introductionmentioning

confidence: 99%

Solution for Ill-Posed Inverse Kinematics of Robot Arm by Network Inversion

Ogawa

Kanada

2010

Journal of Robotics

View full text Add to dashboard Cite

In the context of controlling a robot arm with multiple joints, the method of estimating the joint angles from the given end-effector coordinates is called inverse kinematics, which is a type of inverse problems. Network inversion has been proposed as a method for solving inverse problems by using a multilayer neural network. In this paper, network inversion is introduced as a method to solve the inverse kinematics problem of a robot arm with multiple joints, where the joint angles are estimated from the given end-effector coordinates. In general, inverse problems are affected by ill-posedness, which implies that the existence, uniqueness, and stability of their solutions are not guaranteed. In this paper, we show the effectiveness of applying network inversion with regularization, by which ill-posedness can be reduced, to the ill-posed inverse kinematics of an actual robot arm with multiple joints.

show abstract

Neural network based solution to inverse problems

Cited by 10 publications

References 10 publications

Deep neural-network based optimization for the design of a multi-element surface magnet for MRI applications

Deep neural-network based optimization for the design of a multi-element surface magnet for MRI applications

A novel inverse dynamic model for a magnetorheological damper based on network inversion

Solution for Ill-Posed Inverse Kinematics of Robot Arm by Network Inversion

Contact Info

Product

Resources

About