Optimization of Molecular Characteristics via Machine Learning Based on Continuous Representation of Molecules

We demonstrate an automatic materials design method using continuous representation of molecule and its atomic arrangement via a neural network algorithm. This method is applied to optimizing and predicting the HOMO-LUMO gap within the molecules composed of carbon, oxygen, nitrogen, fluorine, and hydrogen. Adopting the Quantum Machine 9 (QM9) dataset as a training dataset for the molecules, we first established a continuous representation of molecules in a latent space, then predicted molecules that have target values of the HOMO-LUMO gap. In the gap maximization calculation, the CF4 with the largest gap value in the QM9 dataset was automatically found despite there is no a priori data for the gap. In the case of a target gap value of 0.10 hartree, we found a new molecule whose gap value is closer to 0.10 hartree than any other molecules in the QM9 dataset.

Download Full-text

Multimodal Few-Shot Learning for Gait Recognition

Applied Sciences ◽

10.3390/app10217619 ◽

2020 ◽

Vol 10 (21) ◽

pp. 7619

Author(s):

Jucheol Moon ◽

Nhat Anh Le ◽

Nelson Hebert Minaya ◽

Sang-Il Choi

Keyword(s):

Neural Network ◽

Time Series ◽

Gait Recognition ◽

Gait Pattern ◽

Recognition Problem ◽

Human Gait ◽

Training Dataset ◽

Support Vector ◽

Open Set ◽

Latent Space

A person’s gait is a behavioral trait that is uniquely associated with each individual and can be used to recognize the person. As information about the human gait can be captured by wearable devices, a few studies have led to the proposal of methods to process gait information for identification purposes. Despite recent advances in gait recognition, an open set gait recognition problem presents challenges to current approaches. To address the open set gait recognition problem, a system should be able to deal with unseen subjects who have not included in the training dataset. In this paper, we propose a system that learns a mapping from a multimodal time series collected using insole to a latent (embedding vector) space to address the open set gait recognition problem. The distance between two embedding vectors in the latent space corresponds to the similarity between two multimodal time series. Using the characteristics of the human gait pattern, multimodal time series are sliced into unit steps. The system maps unit steps to embedding vectors using an ensemble consisting of a convolutional neural network and a recurrent neural network. To recognize each individual, the system learns a decision function using a one-class support vector machine from a few embedding vectors of the person in the latent space, then the system determines whether an unknown unit step is recognized as belonging to a known individual. Our experiments demonstrate that the proposed framework recognizes individuals with high accuracy regardless they have been registered or not. If we could have an environment in which all people would be wearing the insole, the framework would be used for user verification widely.

Download Full-text

Controller Design Based on Mechatronic Model

Applied Mechanics and Materials ◽

10.4028/www.scientific.net/amm.526.351 ◽

2014 ◽

Vol 526 ◽

pp. 351-356

Author(s):

Li Xi Yue ◽

Jian Hui Zhou ◽

Yan Nan Lu ◽

Chong Chong Ji ◽

Zhi Yong Yu

Keyword(s):

Neural Network ◽

High Speed ◽

Controller Design ◽

Design Method ◽

Nonlinear Modeling ◽

Digital Design ◽

The Neural Network ◽

Key Issues ◽

Neural Network Algorithm ◽

Computed Torque

The dissertation deals with some key issues relevant to the controller design and digital design method for a newly patented high-speed parallel manipulator. Meanwhile, a Virtual Prototyping based co-simulation platform is also established according to the ADAMS and Matlab/Simulink software. In order to promote the ability that the manipulator traces the prescribed trajectory, a model based computed torque controller is described in detail, and a neural network algorithm is also used to optimize controller parameters real-timely under the consideration of systematic nonlinear, modeling error and outer disturbance. The neural network based computed torque controller increases the robustness of system dramatically.

Download Full-text

Application of Artificial Neural Networks to Concrete Pavement Joint Evaluation

Transportation Research Record Journal of the Transportation Research Board ◽

10.1177/0361198196154000108 ◽

1996 ◽

Vol 1540 (1) ◽

pp. 56-64 ◽

Cited By ~ 2

Author(s):

Anastasios M. Ioannides ◽

Don R. Alexander ◽

Michael I. Hammons ◽

Craig M. Davis

Keyword(s):

Neural Network ◽

Dimensional Analysis ◽

Load Transfer ◽

A Priori ◽

The United States ◽

Concrete Pavement ◽

Data Generation ◽

The Neural Network ◽

Neural Network Algorithm ◽

Joint Evaluation

Application of the principles of dimensional analysis has recently led to the development of a robust method for assessing the deflection and stress load transfer efficiencies of concrete pavement joints and for backcalculating joint parameters. The new method eliminates the need to make a priori assumptions since pertinent inputs can now be experimentally determined using the falling weight deflectometer. A data base has been generated using numerical integration of Westergaard-type integrals and has been used to train a backpropagation neural network algorithm for joint evaluation. The resulting computer program is simple, efficient, and precise and can be used on site for immediate results. Its predictions are verified by comparisons with closed-form and finite-element solutions pertaining to data collected at three major civilian airports in the United States, including the new Denver International Airport. Also discussed is the role of dimensional analysis in the generation of the training set for a neural network. It is demonstrated that significant savings can be achieved through reduction of the dimensionality of the problem, which could be reinvested in broadening the range of applicability of the neural network. Comparison of neural network predictions with those from conventional regression analysis and from direct interpolation illustrates the benefits of data generation on the basis of fundamental principles of mechanics.

Download Full-text

A Novel Active Semisupervised Convolutional Neural Network Algorithm for SAR Image Recognition

Computational Intelligence and Neuroscience ◽

10.1155/2017/3105053 ◽

2017 ◽

Vol 2017 ◽

pp. 1-8 ◽

Cited By ~ 4

Author(s):

Fei Gao ◽

Zhenyu Yue ◽

Jun Wang ◽

Jinping Sun ◽

Erfu Yang ◽

...

Keyword(s):

Neural Network ◽

Convolutional Neural Network ◽

Synthetic Aperture ◽

Training Dataset ◽

Training Phase ◽

Sar Image ◽

Initial Training ◽

Neural Network Algorithm ◽

Class Probability ◽

Probability Information

Convolutional neural network (CNN) can be applied in synthetic aperture radar (SAR) object recognition for achieving good performance. However, it requires a large number of the labelled samples in its training phase, and therefore its performance could decrease dramatically when the labelled samples are insufficient. To solve this problem, in this paper, we present a novel active semisupervised CNN algorithm. First, the active learning is used to query the most informative and reliable samples in the unlabelled samples to extend the initial training dataset. Next, a semisupervised method is developed by adding a new regularization term into the loss function of CNN. As a result, the class probability information contained in the unlabelled samples can be maximally utilized. The experimental results on the MSTAR database demonstrate the effectiveness of the proposed algorithm despite the lack of the initial labelled samples.

Download Full-text

A practical demonstration on AMSU retrieval precision for upper tropospheric humidity by a non-linear multi-channel regression method

Atmospheric Chemistry and Physics ◽

10.5194/acp-5-451-2005 ◽

2005 ◽

Vol 5 (2) ◽

pp. 451-459 ◽

Cited By ~ 4

Author(s):

C. Jiménez ◽

P. Eriksson ◽

V. O. John ◽

S. A. Buehler

Keyword(s):

Neural Network ◽

Relative Humidity ◽

Single Channel ◽

Weighting Function ◽

Traditional Approach ◽

Training Dataset ◽

The Neural Network ◽

Radiosonde Station ◽

Neural Network Algorithm ◽

The Impact

Abstract. A neural network algorithm inverting selected channels from the Advance Microwave Sounding Unit instruments AMSU-A and AMSU-B was applied to retrieve layer averaged relative humidity. The neural network was trained with a global synthetic dataset representing clear-sky conditions. A precision of around 6% was obtained when retrieving global simulated radiances, the precision deteriorated less than 1% when real mid-latitude AMSU radiances were inverted and compared with co-located data from a radiosonde station. The 6% precision outperforms by 1% the reported precision estimate from a linear single-channel regression between radiance and weighting function averaged relative humidity, the more traditional approach to exploit AMSU data. Added advantages are not only a better precision; the AMSU-B humidity information is more optimally exploited by including temperature information from AMSU-A channels; and the layer averaged humidity is a more physical quantity than the weighted humidity, for comparison with other datasets. The training dataset proved adequate for inverting real radiances from a mid-latitude site, but it is limited by not considering the impact of clouds or surface emissivity changes, and further work is needed in this direction for further validation of the precision estimates.

Download Full-text

A practical demonstration on AMSU retrieval precision for upper tropospheric humidity by a non-linear multi-channel regression method

Atmospheric Chemistry and Physics Discussions ◽

10.5194/acpd-4-7487-2004 ◽

2004 ◽

Vol 4 (6) ◽

pp. 7487-7511 ◽

Cited By ~ 1

Author(s):

C. Jiménez ◽

P. Eriksson ◽

V. O. John ◽

S. A. Buehler

Keyword(s):

Neural Network ◽

Relative Humidity ◽

Single Channel ◽

Weighting Function ◽

Traditional Approach ◽

Training Dataset ◽

The Neural Network ◽

Radiosonde Station ◽

Neural Network Algorithm ◽

The Impact

Abstract. A neural network algorithm inverting selected channels from the AMSU-A and AMSU-B instruments was applied to retrieve layer averaged relative humidity. The neural network was trained with a global synthetic dataset representing clear-sky conditions. A precision of around 6% was obtained when retrieving global simulated radiances, the precision deteriorated less than 1% when real mid-latitude AMSU radiances were inverted and compared with co-located data from a radiosonde station. The 6% precision outperforms by 1% the reported precision estimate from a linear single-channel regression between radiance and weighting function averaged relative humidity, the more traditional approach to exploit AMSU data. Added advantages are not only a better precision; the AMSU-B humidity information is more optimally exploited by including temperature information from AMSU-A channels; and the layer averaged humidity is a more physical quantity than the weighted humidity, for comparison with other datasets. The training dataset proved adequate for inverting real radiances from a mid-latitude site, but it is limited by not considering the impact of clouds or surface emissivity changes, and further work is needed in this direction for further validation of the precision estimates.

Download Full-text

Research on Power Amplifier for AMB Based on DSP

Applied Mechanics and Materials ◽

10.4028/www.scientific.net/amm.150.217 ◽

2012 ◽

Vol 150 ◽

pp. 217-220

Author(s):

Jian Sheng Zhang ◽

Da Jun Jiang

Keyword(s):

Neural Network ◽

Control System ◽

Power Amplifier ◽

Electromagnetic Force ◽

Design Method ◽

Rbf Neural Network ◽

Magnetic Bearing ◽

Active Magnetic Bearing ◽

System A ◽

Neural Network Algorithm

Active Magnetic Bearing (AMB), which supports freely a rotor by using controllable electromagnetic force, have a lot of advantages such as adjustable stiffness and damp that the traditional bearings can’t compare with. Power Amplifier is an important part of AMB Control System. It can magnify or convert the control signal which can’t drive the electromagnetic iron actuators directly into a power signal. Hence, the performance of power amplifier plays a crucial role in the technique capability of control system. A design method of the PWM Power Amplifier was introduced through the research of the power Amplifier, and hence the CNC fault diagnoses are realized by RBF neural network algorithm and program.

Download Full-text