Ensemble Learning Approaches in Speech Recognition

Zhao, Yunxin; Xue, Jian; Chen, Xin

doi:10.1007/978-1-4939-1456-2_5

Cited by 6 publications

(4 citation statements)

References 75 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Previous studies [18,19] show that an ensemble of models outperforms the single models obviously. However, it is cumbersome to deploy the ensemble to make predictions in real-world systems.…”

Section: Introductionmentioning

confidence: 99%

Distilling Knowledge from an Ensemble of Models for Punctuation Prediction

Tao

Wen

et al. 2017

Interspeech 2017

View full text Add to dashboard Cite

This paper proposes an approach to distill knowledge from an ensemble of models to a single deep neural network (DNN) student model for punctuation prediction. This approach makes the DNN student model mimic the behavior of the ensemble. The ensemble consists of three single models. Kullback-Leibler (KL) divergence is used to minimize the difference between the output distribution of the DNN student model and the behavior of the ensemble. Experimental results on English IWSLT2011 dataset show that the ensemble outperforms the previous state-of-the-art model by up to 4.0% absolute in overall F 1-score. The DNN student model also achieves up to 13.4% absolute overall F 1-score improvement over the conventionally-trained baseline models.

show abstract

Section: Introductionmentioning

confidence: 99%

Distilling Knowledge from an Ensemble of Models for Punctuation Prediction

Tao

Wen

et al. 2017

Interspeech 2017

View full text Add to dashboard Cite

show abstract

“…These networks possess a remarkable ability to effectively acquire new knowledge, to the point that the retention of this information seems to happen effortlessly and lasts for an extended period. Some of the commonly used LSTM models for multi-temporal projections are stacked LSTM [16], vanilla LSTM, bidirectional LSTM [17], and CNN-LSTM [18,19]. The stacked LSTM design is characterized by its use of many linked LSTM layers.…”

Section: Introductionmentioning

confidence: 99%

Testing the Performance of LSTM and ARIMA Models for In-Season Forecasting of Canopy Cover (CC) in Cotton Crops

Dhal,

Kalafatis,

Braga-Neto

et al. 2024

Remote Sensing

View full text Add to dashboard Cite

Cotton (Gossypium spp.), a crucial cash crop in the United States, requires the constant monitoring of growth parameters for informed decision-making. Recently, forecasting models have gained prominence for predicting canopy indicators, aiding in-season planning and management decisions to optimize cotton production. This study employed unmanned aerial system (UAS) technology to collect canopy cover (CC) data from a 40-hectare cotton field in Driscoll, Texas, in 2020 and 2021. Long short-term memory (LSTM) models, trained using 2020 data, were subsequently applied to forecast the CC values for 2021. These models were compared with real-time auto-regressive integrated moving average (ARIMA) models to assess their effectiveness in predicting the CC values up to 14 days in advance, starting from the 28th day after crop emergence. The results showed that multiple-input multi-step output LSTM models achieved higher accuracy in predicting the in-season CC values during the early growth stages (up to the 56th day), with an average testing RMSE of 3.86, significantly lower than other single-input LSTM models. Conversely, when sufficient testing data are available, single-input stacked-LSTM models demonstrated precision in CC predictions for later stages, achieving an average RMSE of 3.06. These findings highlight the potential of LSTM models for in-season CC forecasting, facilitating effective management strategies in cotton production.

show abstract

“…Machine learning offers several classification techniques for punctuation prediction, such as Support Vector Machines (SVM) (Akita et al, 2006), Neural Networks (NN) (Lample et al, 2016; Ren et al, 2004; Vadapalli & Gangashetty, 2016), Classification And Regression Trees (CART) (Sarkar & Rao, 2015), Adaptive Boosting (AdaBoost) (Kolář & Lamel, 2012), Conditional Random Forests (CRF) (Wang et al, 2012). In recent times, ensemble‐based (Yi et al, 2017; Zhao et al, 2015) (which combine two or more of the above base classifiers) and attention‐based (Yi & Tao, 2019) models have been found to perform better especially for the task of punctuation prediction. Apart from exploring these classifier models for punctuation prediction in Tamil and Hindi texts, the current work proposes a modified Feature‐weighted AdaBoost (FAda) classifier which improves the prediction accuracy of the Adaboost classifier.…”

Section: Introductionmentioning

confidence: 99%

Feature‐weightedAdaBoostclassifier for punctuation prediction in Tamil and HindiNLPsystems

2021

View full text Add to dashboard Cite

Punctuation marks play a vital role in text representation and interpretation, and are useful in enhancing the performance of modern Natural Language Processing (NLP) systems such as voice input typing aids, machine translation, and speech synthesis systems. Punctuation marks, except period, are inherently not available in Indian languages such as Tamil and Hindi. However, some modern forms of writing such as news articles, blogs, stories, and so forth, incorporate user‐defined punctuation marks in these languages. The current work proposes an automatic punctuation prediction system for texts in Tamil and Hindi using classification approach, where punctuation prediction is considered as a multi‐class classification problem. Word‐level text features are chosen and are analysed to validate their language‐dependency and significance towards punctuation prediction. A Feature‐weighted AdaBoost (FAda) classifier is proposed that defines a novel boosting factor to adjust the hypothesis weight of the weak classifiers, hence reducing the number of false classifications. It is observed that the proposed classifier outperforms the other classification techniques such as, AdaBoost, SVM, CART, CRF, and Bi‐LSTM by a maximum difference of 50% and 16% in the macro F1‐scores for Tamil and Hindi texts, respectively. The proposed classifier performs on par with the attention‐based classifier for both Tamil and Hindi texts. Further, as a proof of concept, the proposed punctuation prediction system is applied to voice keyboard, machine translation, and speech synthesis systems, to validate the effect of the punctuation marks on the performance of these Natural Language Processing (NLP) systems.

show abstract

Ensemble Learning Approaches in Speech Recognition

Cited by 6 publications

References 75 publications

Distilling Knowledge from an Ensemble of Models for Punctuation Prediction

Distilling Knowledge from an Ensemble of Models for Punctuation Prediction

Testing the Performance of LSTM and ARIMA Models for In-Season Forecasting of Canopy Cover (CC) in Cotton Crops

Feature‐weightedAdaBoostclassifier for punctuation prediction in Tamil and HindiNLPsystems

Contact Info

Product

Resources

About