scholarly journals ASPECTS OF TEMPERATURE TAKING INTO ACCOUNT TO INCREASE THE ACCURACY OF SHORT-TERM FORECASTING OF NODE LOADS

2021 ◽  
Author(s):  
P. Shymaniuk ◽  
V Miroshnyk ◽  
I. Blinov ◽  
P. Chernenko
Keyword(s):  
Neural Network ◽  
Power Systems ◽  
Forecast Accuracy ◽  
Training Sample ◽  
Neural Network Training ◽  
Analysis Method ◽  
Short Term ◽  
Data Distortion ◽  
Network Training ◽  
Short Term Forecasting

The peculiarities of the influence of air temperature data on the accuracy of forecasting of nodal loads in power systems and how the accuracy of such forecasting changes depending on the training sample and its volume are considered. The application of the data analysis method to detect anomalous values ​​and omissions to reduce data distortion and improve forecasting results is considered. A neural network of deep learning of the LSTM type was used for multifactor prediction of nodal loads. To evaluate the effectiveness of the forecast accuracy, various variants of data samples for neural network training are considered.

The Short-Term Forecast Model of Pork Price Based on CNN-GA

2012 ◽  
Vol 628 ◽  
pp. 350-358 ◽  
Author(s):  
Zhe Min Li ◽  
Shi Wei Xu ◽  
Li Guo Cui ◽  
Gan Qiong Li ◽  
Xiao Xia Dong ◽  
...  
Keyword(s):  
Neural Network ◽  
Forecast Accuracy ◽  
Forecast Model ◽  
Basic Structure ◽  
Retail Price ◽  
Practical Significance ◽  
Short Term ◽  
The Neural Network ◽  
High Prediction ◽  
Short Term Forecasting

After analyzing and reviewing the short-term forecasting methods research of pork price at home and abroad, a chaotic neural network model based on genetic algorithm (CNN-GA) was put forward according to the nonlinear characteristics of pork price,which established on the base of the chaotic theory and the neural network technology. Chosen the daily retail price data of the pork (streaky pork) from January 1, 2008 to June 11, 2012,we designed the basic structure of CNN-GA, and thentrainedit in order to attain the trained CNN-GA model. Finally, the trained CNN-GA model was used to predict the 20 days’ (from June 12, 2012 to July 1, 2012) retail price of pork (streaky pork) and then compared the predicted price with the real price to test the model’s forecast accuracy and application ability.The result shows that the model has high prediction precision, good fitting effect and hasan important reference and practical significance for the short-term price forecasting of the pork market.

scholarly journals The training set in the problem of pattern recognition using neural networks

2015 ◽  
Vol 11 (2) ◽  
pp. 94-98
Author(s):  
Vitaly M Tatyankin ◽  
Irina S Dyubko
Keyword(s):  
Neural Network ◽  
Neural Networks ◽  
Pattern Recognition ◽  
Training Sample ◽  
Neural Network Training ◽  
Training Set ◽  
Network Training ◽  
Practical Recommendations

The article discusses approaches to the formation of the training sample in the problem of recognition of monochrome images. It is shown that the variation of the training sample, allows to reduce the error of neural network training. Practical recommendations for the formation of a training sample.

scholarly journals A Geometric Perspective on Information Plane Analysis

Entropy ◽  
2021 ◽  
Vol 23 (6) ◽  
pp. 711
Author(s):  
Mina Basirat ◽  
Bernhard C. Geiger ◽  
Peter M. Roth
Keyword(s):  
Neural Network ◽  
Mutual Information ◽  
Geometric Interpretation ◽  
Neural Network Training ◽  
Neural Network Learning ◽  
Network Learning ◽  
Plane Analysis ◽  
Network Training ◽  
Hidden Layer ◽  
The Impact

Information plane analysis, describing the mutual information between the input and a hidden layer and between a hidden layer and the target over time, has recently been proposed to analyze the training of neural networks. Since the activations of a hidden layer are typically continuous-valued, this mutual information cannot be computed analytically and must thus be estimated, resulting in apparently inconsistent or even contradicting results in the literature. The goal of this paper is to demonstrate how information plane analysis can still be a valuable tool for analyzing neural network training. To this end, we complement the prevailing binning estimator for mutual information with a geometric interpretation. With this geometric interpretation in mind, we evaluate the impact of regularization and interpret phenomena such as underfitting and overfitting. In addition, we investigate neural network learning in the presence of noisy data and noisy labels.

Sign in / Sign up

Export Citation Format

Share Document