scholarly journals Reusing Source Task Knowledge via Transfer Approximator in Reinforcement Transfer Learning

Symmetry ◽  
2018 ◽  
Vol 11 (1) ◽  
pp. 25 ◽  
Author(s):  
Qiao Cheng ◽  
Xiangke Wang ◽  
Yifeng Niu ◽  
Lincheng Shen
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Transfer Learning ◽  
Mapping Rule ◽  
Learning Methods ◽  
Target Task ◽  
Artificial Neural ◽  
Artificial Neural Network Ann ◽  
The One

Transfer Learning (TL) has received a great deal of attention because of its ability to speed up Reinforcement Learning (RL) by reusing learned knowledge from other tasks. This paper proposes a new transfer learning framework, referred to as Transfer Learning via Artificial Neural Network Approximator (TL-ANNA). It builds an Artificial Neural Network (ANN) transfer approximator to transfer the related knowledge from the source task into the target task and reuses the transferred knowledge with a Probabilistic Policy Reuse (PPR) scheme. Specifically, the transfer approximator maps the state of the target task symmetrically to states of the source task with a certain mapping rule, and activates the related knowledge (components of the action-value function) of the source task as the input of the ANNs; it then predicts the quality of the actions in the target task with the ANNs. The target learner uses the PPR scheme to bias the RL with the suggested action from the transfer approximator. In this way, the transfer approximator builds a symmetric knowledge path between the target task and the source task. In addition, two mapping rules for the transfer approximator are designed, namely, Full Mapping Rule and Group Mapping Rule. Experiments performed on the RoboCup soccer Keepaway task verified that the proposed transfer learning methods outperform two other transfer learning methods in both jumpstart and time to threshold metrics and are more robust to the quality of source knowledge. In addition, the TL-ANNA with the group mapping rule exhibits slightly worse performance than the one with the full mapping rule, but with less computation and space cost when appropriate grouping method is used.

Adaptive Power-Saving Mechanism for VoIP Over WiMAX Based on Artificial Neural Network

2019 ◽  
pp. 158-177
Author(s):  
Tamer Emara
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Power Saving ◽  
Ieee 802.16 ◽  
Feed Forward Neural Network ◽  
Artificial Neural ◽  
Adaptive Power ◽  
Artificial Neural Network Ann ◽  
Hidden Layer

The IEEE 802.16 system offers power-saving class type II as a power-saving algorithm for real-time services such as voice over internet protocol (VoIP) service. However, it doesn't take into account the silent periods of VoIP conversation. This chapter proposes a power conservation algorithm based on artificial neural network (ANN-VPSM) that can be applied to VoIP service over WiMAX systems. Artificial intelligent model using feed forward neural network with a single hidden layer has been developed to predict the mutual silent period that used to determine the sleep period for power saving class mode in IEEE 802.16. From the implication of the findings, ANN-VPSM reduces the power consumption during VoIP calls with respect to the quality of services (QoS). Experimental results depict the significant advantages of ANN-VPSM in terms of power saving and quality-of-service (QoS). It shows the power consumed in the mobile station can be reduced up to 3.7% with respect to VoIP quality.

scholarly journals Length of Hospital Stay Prediction at the Admission Stage for Cardiology Patients Using Artificial Neural Network

2016 ◽  
Vol 2016 ◽  
pp. 1-11 ◽  
Author(s):  
Pei-Fang (Jennifer) Tsai ◽  
Po-Chia Chen ◽  
Yen-You Chen ◽  
Hao-Yuan Song ◽  
Hsiu-Mei Lin ◽  
...  
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Coronary Atherosclerosis ◽  
Inpatient Care ◽  
Length Of Hospital Stay ◽  
Ann Models ◽  
Artificial Neural ◽  
Artificial Neural Network Ann ◽  
Christian Hospital

For hospitals’ admission management, the ability to predict length of stay (LOS) as early as in the preadmission stage might be helpful to monitor the quality of inpatient care. This study is to develop artificial neural network (ANN) models to predict LOS for inpatients with one of the three primary diagnoses: coronary atherosclerosis (CAS), heart failure (HF), and acute myocardial infarction (AMI) in a cardiovascular unit in a Christian hospital in Taipei, Taiwan. A total of 2,377 cardiology patients discharged between October 1, 2010, and December 31, 2011, were analyzed. Using ANN or linear regression model was able to predict correctly for 88.07% to 89.95% CAS patients at the predischarge stage and for 88.31% to 91.53% at the preadmission stage. For AMI or HF patients, the accuracy ranged from 64.12% to 66.78% at the predischarge stage and 63.69% to 67.47% at the preadmission stage when a tolerance of 2 days was allowed.

Artificial neural network as a predictive tool for emissions from heavy-duty diesel vehicles in Southern California

2007 ◽  
Vol 8 (4) ◽  
pp. 321-336 ◽  
Author(s):  
N Hashemi ◽  
N. N. Clark
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Oxides Of Nitrogen ◽  
Heavy Duty ◽  
Predictive Tool ◽  
Diesel Vehicles ◽  
Artificial Neural ◽  
Heavy Duty Diesel ◽  
Artificial Neural Network Ann ◽  
The One

An artificial neural network (ANN) was trained on chassis dynamometer data and used to predict the oxides of nitrogen (NO x), carbon dioxide (CO2), hydrocarbons (HC), and carbon monoxide (CO) emitted from heavy-duty diesel vehicles. Axle speed, torque, their derivatives in different time steps, and two novel variables that defined speed variability over 150 seconds were defined as the inputs for the ANN. The novel variables were used to assist in predicting off-cycle emissions. Each species was considered individually as an output of the ANN. The ANN was trained on the Highway cycle and applied to the City/Suburban Heavy Vehicle Route (CSHVR) and Urban Dynamometer Driving Schedule (UDDS) with four different sets of inputs to predict the emissions for these vehicles. The research showed acceptable prediction results for the ANN, even for the one trained with only eight inputs of speed, torque, their first and second derivatives at one second, and two variables related to the speed pattern over the last 150 seconds. However, off-cycle operation (leading to high NO x emissions) was still difficult to model. The results showed an average accuracy of 0.97 for CO2, 0.89 for NO x, 0.70 for CO, and 0.48 for HC over the course of the CSHVR, Highway, and UDDS.

Adaptive Power-Saving Mechanism for VoIP Over WiMAX Based on Artificial Neural Network

2022 ◽  
pp. 471-489
Author(s):  
Tamer Emara
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Power Saving ◽  
Ieee 802.16 ◽  
Feed Forward Neural Network ◽  
Artificial Neural ◽  
Adaptive Power ◽  
Artificial Neural Network Ann ◽  
Hidden Layer

The IEEE 802.16 system offers power-saving class type II as a power-saving algorithm for real-time services such as voice over internet protocol (VoIP) service. However, it doesn't take into account the silent periods of VoIP conversation. This chapter proposes a power conservation algorithm based on artificial neural network (ANN-VPSM) that can be applied to VoIP service over WiMAX systems. Artificial intelligent model using feed forward neural network with a single hidden layer has been developed to predict the mutual silent period that used to determine the sleep period for power saving class mode in IEEE 802.16. From the implication of the findings, ANN-VPSM reduces the power consumption during VoIP calls with respect to the quality of services (QoS). Experimental results depict the significant advantages of ANN-VPSM in terms of power saving and quality-of-service (QoS). It shows the power consumed in the mobile station can be reduced up to 3.7% with respect to VoIP quality.

Prediction of Springback of the One-Axle Rotary Shaping Based on Artificial Neural Network

2013 ◽  
Vol 535-536 ◽  
pp. 318-321
Author(s):  
Xia Jin ◽  
Shi Hong Lu
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Simulation Software ◽  
Forming Process ◽  
Training Samples ◽  
Metal Forming Process ◽  
Artificial Neural ◽  
Artificial Neural Network Ann ◽  
Set Up ◽  
The One

One-axle rotary shaping with the elastic medium (RSEM) is a kind of advanced sheet metal forming process. The research object is the springback of aluminous U-section. The orthogonal method is used to arrange the simulation experiments, the forming and springback of the workpiece are simulated successfully with the Finite Element Simulation software, and The main factors influenced the RSEM are analyzed. The simulation results are used as the training samples of the artificial neural network (ANN), and the ANN prediction model of RSEM process is set up. The prediction results would be tested with the experiment data, and only a little tolerance was existed between the two values. It demonstrated that the combination of orthogonal test, numerical simulation and neural network could effectively predict the springback of RSEM, the design efficiency of process parameters would be improved. It would guide the development of precision forming technology.

Modeling of Moment of Friction under Increasing Load Using Artificial Neural Networks

2008 ◽  
Vol 144 ◽  
pp. 130-135
Author(s):  
Krzysztof Gocman ◽  
Bolesław Giemza ◽  
Tadeusz Kałdoński
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Artificial Neural Network ◽  
Rotational Speed ◽  
Training Algorithms ◽  
Analysis Of Results ◽  
Artificial Neural ◽  
Artificial Neural Network Ann ◽  
Increasing Load

Preliminary results of testing of influence of load and rotational speed on moment of friction are presented in this paper. Tests were carried out under increasing load and within the range of rotational speed of 500 – 1500 rpm. The analysis of results was elaborated and model of moment of friction was developed on the basis of artificial neural network (ANN). Different kind of networks and various training algorithms were applied in order to obtain the best quality of the developed models.

scholarly journals Using an artificial neural network to develop an optimal model of straight punch in boxing and training in punch techniques based on this model and real-time feedback

PLoS ONE ◽  
2021 ◽  
Vol 16 (11) ◽  
pp. e0259457
Author(s):  
Ilshat Khasanshin ◽  
Aleksey Osipov
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Highly Qualified ◽  
Optimal Model ◽  
Data Set ◽  
Punch Force ◽  
Angular Velocities ◽  
Artificial Neural ◽  
Artificial Neural Network Ann

The work was aimed to develop an optimal model of a straight punch in boxing based on an artificial neural network (ANN) in the form of a multilayer perceptron, as well as to develop a technique for improving the technique of punches in boxing based on feedback, when each punch delivered by a boxer was compared with the optimal model. The architecture of the neural network optimal punch model included an input layer of 600 nodes—the values of absolute accelerations and angular velocities, four hidden ones, as well as a binary output layer (the best and not the best punch). To measure accelerations and angular velocities, inertial measuring devices were attached to the boxers’ wrists. Highly qualified participated in the data set for the development of the optimal model. The best punches were chosen according to the criteria of strength and speed. The punch force was determined using a boxing pad with the function of measuring the punch force. In order to be able to compare punches, a unified parameter was developed, called the punch quality, which is equal to the product of the effective force and the punch speed. To study the effects of biofeedback, the boxing pads were equipped with five LEDs. The more LEDs were turned on, the more the punch corresponded to the optimal model. As a result of the study, an almost linear relationship was found between the quality of the punch of entry-level boxers and the optimal model. The use of feedback allowed for an increase in the quality of punches from 11 to 25%, which is on average twice as high as in the group where the feedback method was not used. Studies have shown that it is possible to develop an optimal punch model. According to the degree of compliance with this model, you can evaluate and train boxers in the technique.

Artificial Neural Network Modeling for Electrical Discharge Machining Parameters

2014 ◽  
pp. 281-302
Author(s):  
Raja Das ◽  
M. K. Pradhan
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Electrical Discharge ◽  
Electrical Discharge Machining ◽  
Network Models ◽  
Neural Network Models ◽  
Copper Electrodes ◽  
Artificial Neural ◽  
Artificial Neural Network Ann

The objective of the chapter is to present the application of Artificial Neural Network (ANN) modelling of the Electrical Discharge Machining (EDM) process. It establishes the best ANN model by comparing the prediction from different models under the effect of process parameters. In EDM, the motivation is frequently to get better Material Removal Rate (MRR) with fulfilling better surface quality of machined components. The vital requirements are as small a radial overcut with minimal tool wear rate. The quality of a machined surface is very important to fulfilling the growing demands of higher component performance, durability, and reliability. To improve the reliability of the machine component, it is necessary to have in depth knowledge of the effect of parameters on the aforesaid responses of the components. An extensive chain of experiments has been conducted over a wide range of input parameters, using the full factorial design. More than 150 experiments have been conducted on AISI D2 work piece materials using copper electrodes to get the data for training and testing. The additional experiments were obtained to validate the model predictions. The performance of three neural network models is discussed in the evaluation of the generalization ability of the trained neural network. It was observed that the artificial neural network models could predict the process performance with reasonable accuracy, under varying machining conditions.

scholarly journals Internship Effect Prediction for Physical Education Majors Based on Artificial Neural Network

2021 ◽  
Vol 16 (24) ◽  
pp. 165-176
Author(s):  
Bo Yang
Keyword(s):  
Neural Network ◽  
Artificial Neural Network ◽  
Physical Education ◽  
Prediction Model ◽  
Evaluation Index System ◽  
Job Competencies ◽  
Physical Education Majors ◽  
Artificial Neural ◽  
Artificial Neural Network Ann

Professional internship offers college students a golden chance to apply their theoretical knowledge to practice. Through internship, physical education (PE) majors can match the professional knowledge and skills learned at school with the competencies required by actual jobs. The relevant studies at home and abroad mainly attempt to improve the internship effect. This paper explores the influence of the diversity of job competencies on the internship effect of PE majors, and establishes a prediction model based on artificial neural network (ANN). Firstly, an evaluation index system (EIS) was constructed for the internship quality of PE majors, and a table was prepared for four types of internship jobs for PE majors, as well as their core competences. Then, the sample data for quality evaluation of PE majors’ internship were preprocessed and subjected to feature extraction, in the light of their sequential property. After that, a prediction model was proposed for the internship quality of PE majors, along with its optimization algorithm. The proposed model was proved effective through experiments.

Sign in / Sign up

Export Citation Format

Share Document