The Experts below are selected from a list of 213 Experts worldwide ranked by ideXlab platform
Martin Riedmiller - One of the best experts on this subject based on the ideXlab platform.
-
learn to swing up and balance a Real Pole based on raw visual input data
International Conference on Neural Information Processing, 2012Co-Authors: Jan Mattner, Sascha Lange, Martin RiedmillerAbstract:For the challenging Pole balancing task we propose a system which uses raw visual input data for reinforcement learning to evolve a control strategy. Therefore we use a neural network --- a deep autoencoder --- to encode the camera images and thus the system states in a low dimensional feature space. The system is compared to controllers that work directly on the motor sensor data. We show that the performances of both systems are settled in the same order of magnitude.
-
neural reinforcement learning to swing up and balance a Real Pole
Systems Man and Cybernetics, 2005Co-Authors: Martin RiedmillerAbstract:This paper proposes a neural network based reinforcement learning controller that is able to learn control policies in a highly data efficient manner. This allows to apply reinforcement learning directly to Real plants -neither a transition model nor a simulation model of the plant is needed for training. The only training information provided to the controller are transition experiences collected from interactions with the Real plant. By storing these transition experiences explicitly, they can be reconsidered for updating the neural Q-function in every training step. This results in a stable learning process of a neural Q-value function. The algorithm is applied to learn the highly nonlinear and noisy task of swinging-up and balancing a Real inverted pendulum. The amount of Real time interaction needed to learn a highly effective policy from scratch was less than 14 minutes.
-
SMC - Neural reinforcement learning to swing-up and balance a Real Pole
2005 IEEE International Conference on Systems Man and Cybernetics, 1Co-Authors: Martin RiedmillerAbstract:This paper proposes a neural network based reinforcement learning controller that is able to learn control policies in a highly data efficient manner. This allows to apply reinforcement learning directly to Real plants -neither a transition model nor a simulation model of the plant is needed for training. The only training information provided to the controller are transition experiences collected from interactions with the Real plant. By storing these transition experiences explicitly, they can be reconsidered for updating the neural Q-function in every training step. This results in a stable learning process of a neural Q-value function. The algorithm is applied to learn the highly nonlinear and noisy task of swinging-up and balancing a Real inverted pendulum. The amount of Real time interaction needed to learn a highly effective policy from scratch was less than 14 minutes.
Silvio P. Sorella - One of the best experts on this subject based on the ideXlab platform.
-
Confinement and dynamical chiral symmetry breaking in a non-perturbative renormalizable quark model
Annals of Physics, 2016Co-Authors: David Dudal, M. S. Guimaraes, L. F. Palhares, Silvio P. SorellaAbstract:Abstract Inspired by the construction of the Gribov–Zwanziger action in the Landau gauge, we introduce a quark model exhibiting both confinement and chiral symmetry aspects. An important feature is the incorporation of spontaneous chiral symmetry breaking in a renormalizable fashion. The quark propagator in the condensed vacuum turns out to be of a confining type. Besides a Real Pole, it exhibits complex conjugate Poles. The resulting spectral form is explicitly shown to violate positivity, indicative of its unphysical character. Moreover, the ensuing quark mass function fits well to existing lattice data. To further validate the physical nature of the model, we identify a massless pseudoscalar (i.e. a pion) in the chiral limit and present estimates for the ρ meson mass and decay constant.
-
Confinement and dynamical chiral symmetry breaking in a non-perturbative renormalizable quark model
'Elsevier BV', 2016Co-Authors: Dudal David, L. F. Palhares, Guimaraes Ms, Silvio P. SorellaAbstract:© 2015 Elsevier Inc. Inspired by the construction of the Gribov-Zwanziger action in the Landau gauge, we introduce a quark model exhibiting both confinement and chiral symmetry aspects. An important feature is the incorporation of spontaneous chiral symmetry breaking in a renormalizable fashion. The quark propagator in the condensed vacuum turns out to be of a confining type. Besides a Real Pole, it exhibits complex conjugate Poles. The resulting spectral form is explicitly shown to violate positivity, indicative of its unphysical character. Moreover, the ensuing quark mass function fits well to existing lattice data. To further validate the physical nature of the model, we identify a massless pseudoscalar (i.e. a pion) in the chiral limit and present estimates for the ρ meson mass and decay constant.publisher: Elsevier articletitle: Confinement and dynamical chiral symmetry breaking in a non-perturbative renormalizable quark model journaltitle: Annals of Physics articlelink: http://dx.doi.org/10.1016/j.aop.2015.12.003 content_type: article copyright: Copyright © 2015 Elsevier Inc. All rights reserved.status: publishe
Jesús Blázquez - One of the best experts on this subject based on the ideXlab platform.
-
In situ surveillance of capacitive pressure transmitter dynamic Poles
Nuclear Technology, 2012Co-Authors: M. Balbás, A. García-berrocal, C. Montalvo, Jesús BlázquezAbstract:In this work, a methodology is proposed to find the dynamic Poles of a capacitive pressure transmitter in order to enhance and extend the online surveillance of this type of sensor based on the response time measurement by applying noise analysis techniques and the dynamic data system procedure. Several measurements taken from a pressurized water reactor have been analyzed. The methodology proposes an autoregressive fit whose order is determined by the sensor dynamic Poles. Nevertheless, the signals that have been analyzed could not be filtered properly in order to remove the plant noise; thus, this was considered as an additional pair of complex conjugate Poles. With this methodology we have come up with the numerical value of the sensor second Real Pole in spite of its low influence on the sensor dynamic response. This opens up a more accurate online sensor surveillance since the previous methods were achieved by considering one Real Pole only.
-
Pressure transmitter surveillance : the dominant Real Pole case
Progress in Nuclear Energy, 1995Co-Authors: Jesús Blázquez, Jesús BallestrínAbstract:Abstract There are about 500 pressure transmitters in a Nuclear Power Plant. Due to Safety requirements, some of them must be specially surveilled. Sensor response time to a pressure ramp is the usual quantity to be measured. Response time, τ r , reflects the dynamics of the sensor and the sensing line. A Real Pole is due to the inner sensor structure, but the complex Pole stands for the sensing line too. The Real Pole usually is the dominant in most sensors. On line monitoring noise analysis regards simultaneously both, the sensor and the sensing line, but the noise signal contains not only the sensor Poles, but many others coming from the Plant, so must be conditioned previously and the determination of τ r is not free of systematic errors. That is the price to be paid for non disturbing the Plant. When the Real Pole is dominant, the sensing line contribution is negligible, so the on line noise monitoring methods are supported by the laboratory experiments and the Real Pole border in the PSD is properly identified. The mean square frequency results proportional to τ r −1 , so manual techniques are designed for response time surveillance made by non noise Plant's maintenance technicians.
F.a. Mata - One of the best experts on this subject based on the ideXlab platform.
-
Robust stabilization of systems with multiple Real Pole uncertainties
IEEE Transactions on Automatic Control, 1991Co-Authors: J. Amillo, F.a. MataAbstract:The robust stabilization of systems with several uncertain Real Poles is discussed. Necessary and sufficient conditions of existence of a controller in terms is obtained of an interpolation problem are given. The interpolating function is obtained directly for a particular case. Two examples are shown for the synthesis of a controller. An explicit solution of the controller for plants without zeros in the right hand plane (RHP) and without known Poles in the RHP is given. This shows the practical utility of the method to the design of robust stabilizers. >
Jesús Ballestrín - One of the best experts on this subject based on the ideXlab platform.
-
Pressure transmitter surveillance : the dominant Real Pole case
Progress in Nuclear Energy, 1995Co-Authors: Jesús Blázquez, Jesús BallestrínAbstract:Abstract There are about 500 pressure transmitters in a Nuclear Power Plant. Due to Safety requirements, some of them must be specially surveilled. Sensor response time to a pressure ramp is the usual quantity to be measured. Response time, τ r , reflects the dynamics of the sensor and the sensing line. A Real Pole is due to the inner sensor structure, but the complex Pole stands for the sensing line too. The Real Pole usually is the dominant in most sensors. On line monitoring noise analysis regards simultaneously both, the sensor and the sensing line, but the noise signal contains not only the sensor Poles, but many others coming from the Plant, so must be conditioned previously and the determination of τ r is not free of systematic errors. That is the price to be paid for non disturbing the Plant. When the Real Pole is dominant, the sensing line contribution is negligible, so the on line noise monitoring methods are supported by the laboratory experiments and the Real Pole border in the PSD is properly identified. The mean square frequency results proportional to τ r −1 , so manual techniques are designed for response time surveillance made by non noise Plant's maintenance technicians.
