LSTM-Based Language Models for Spontaneous Speech Recognition

Medennikov, Ivan; Bulusheva, Anna

doi:10.1007/978-3-319-43958-7_56

Cited by 16 publications

(6 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This effectively also increases the performance of the model on the unseen results (Wan et al 2013). Dropout implies that a unit is temporarily dropped from the network, for all its coming and outgoing links (Medennikov and Bulusheva 2016). All elements of an output layer are stored with probability p, alternatively set to 0 with probability (1 -p).…”

Section: 𝑋 (11)mentioning

confidence: 99%

Electricity demand and price forecasting model for sustainable smart grid using comprehensive long short term memory

Fatema

Kong

Fang

2021

International Journal of Sustainable Engineering

View full text Add to dashboard Cite

This paper proposes an electricity demand and price forecast model of the smart city large datasets using a single comprehensive Long Short-Term Memory (LSTM) based on a sequence-to-sequence network. Real electricity market data from the Australian Energy Market Operator (AEMO) is used to validate the effectiveness of the proposed model. Several simulations with different configurations are executed on actual data to produce reliable results. The validation results indicate that the devised model is a better option to forecast the electricity demand and price with an acceptably smaller error. A comparison of the proposed model is also provided with a few existing models, Support Vector Machine (SVM), Regression Tree (RT), and Neural Nonlinear Autoregressive network with Exogenous variables (NARX). Compared to SVM, RT, and NARX, the performance indices, Root Mean Square Error (RMSE) of the proposed forecasting model has been improved by 11.25%, 20%, and 33.5% respectively considering demand, and by 12.8%, 14.5%, and 47% respectively considering the price; similarly, the Mean Absolute Error (MAE) has been improved by 14%, 22.5%, and 32.5% respectively considering demand, and by 8.4%, 21% and 61% respectively considering price. Additionally, the proposed model can produce reliable forecast results without large historical datasets.

show abstract

Section: 𝑋 (11)mentioning

confidence: 99%

Electricity demand and price forecasting model for sustainable smart grid using comprehensive long short term memory

Fatema

Kong

Fang

2021

International Journal of Sustainable Engineering

View full text Add to dashboard Cite

show abstract

“…x − x min x max − x min (11) where x * is the normalized value, x max is the maximum value in the sample data, x min is the minimum value in the sample data. 800 sets of monitoring data from March 1, 2015 to May 9, 2017 are used as training samples, and 184 sets of monitoring data from May 10, 2017 to November 10, 2017 are used as test samples.…”

Section: Data Preprocessingmentioning

confidence: 99%

“…Long short-term memory (LSTM) network is an effective nonlinear recurrent neural network, which is suitable for language model [9], machine translation [10], speech recognition [11] and other fields. However, although LSTM network model can fully reflect the long-term historical process in input time series data, it cannot excavate the correlation and inherent laws between discontinuous data.…”

Section: Introductionmentioning

confidence: 99%

Research on Winding Temperature Prediction of UHV Transformer Based on Convolutional Long Short-Term Memory Network

Lin

Miao

Xiao

et al. 2021

Lecture Notes in Electrical Engineering

View full text Add to dashboard Cite

In order to improve the prediction accuracy of ultra-high voltage (UHV) transformer winding temperature, a hybrid model based on convolutional neural network (CNN) and long short-term memory (LSTM) network is proposed. This method uses a large number of historical monitoring data, including data of winding temperature, top oil temperature, dissolved gas in oil, environmental data and so on. Combined with the characteristics of CNN and LSTM networks, CNN convolution layer and pooling layer are used to extract relevant data information to generate feature vectors, then the feature vectors are used as input data to predict winding temperature through LSTM network. The example results show that this method is of high prediction accuracy, its prediction accuracy is better than that of single LSTM model, which can effectively improve the prediction accuracy of UHV transformer winding temperature.

show abstract

“…In recent years, the feature learning ability of artificial neural network has become increasingly prominent (Hinton and Salakhutdinov, 2006; Lecun et al , 2015). Among them, deep learning (DL) has achieved good performance in image classification (Russakovsky et al , 2014), image segmentation (Ibtehaz and Rahman, 2019) and natural language processing (Medennikov and Bulusheva, 2016). DL completes the prediction of the results by learning the historical feature information of the numerical simulation results.…”

Section: Introductionmentioning

confidence: 99%

Prediction method of motor magnetic field based on improved Linknet model

Jin

Liu

Yang

et al. 2022

COMPEL

View full text Add to dashboard Cite

Purpose Under the condition of small data set, a prediction model of motor magnetic field is established based on deep learning method. This paper aims to complete the magnetic field prediction quickly and accurately. Design/methodology/approach An improved Linknet model is proposed to predict the motor magnetic field. This is a digital twin technology, which can predict the function values of other points according to the function values of typical sampling points. The results of magnetic field distribution are represented by color images. By predicting the pixels of the image, the corresponding magnetic field distribution is obtained. The model not only considers the correlation between pixels but also retains the spatial information in the original input image and can well learn the mapping relationship between motor structure and magnetic field. Findings The model can speed up the calculation while ensuring the accuracy and has obvious advantages in large-scale calculation and real-time simulation. Originality/value Under the condition of small data set, the model can well learn the mapping relationship between motor structure and magnetic field, so as to complete the magnetic field prediction quickly and accurately. In the future, according to the characteristics of magnetic field distribution, it will lay a foundation for solving the problems of rapid optimization, real-time simulation and physical field control of electrical equipment.

show abstract

LSTM-Based Language Models for Spontaneous Speech Recognition

Cited by 16 publications

References 14 publications

Electricity demand and price forecasting model for sustainable smart grid using comprehensive long short term memory

Electricity demand and price forecasting model for sustainable smart grid using comprehensive long short term memory

Research on Winding Temperature Prediction of UHV Transformer Based on Convolutional Long Short-Term Memory Network

Prediction method of motor magnetic field based on improved Linknet model

Contact Info

Product

Resources

About