Sparse Autoencoder-Based Feature Transfer Learning for Speech Emotion Recognition

By Jun Deng, Zixing Zhang, Erik Marchi, Björn Schuller. 2013

362 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Signal Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Signal Processing

Citations in the corpus

Paraphrasing Complex Network: Network Compression via Factor Transfer (2018, neutral): “sparsity to an overcomplete autoencoder to learn the size of factor needed by itself (Deng et al., 2013) or add some noise at the input data to get robustness from corrupted data by recovering it using denoising autoencoder (Vincent et al.”
Relational autoencoder for feature extraction (2017, background): “Because of this, au-a toencoders and extensions such as Sparse Autoencoders [22], Denoising Autoencoders [23], Contractive Autoencoders [24] and Variational Autoencoders [25], demonstrate a promising ability to extract meaningful features, especially in image processing and natural language processi”
A Comprehensive Survey on Transfer Learning (2019, background): “Transfer learning has been applied to some recognition tasks such as hand gesture recognition [206], face recognition [207], activity recognition [208], and speech emotion recognition [209].”
A Survey of Clustering With Deep Learning: From the Perspective of Network Architecture (2018, background): “, sparse autoencoder [36] imposes a sparsity constraint on latent coder to obtain a sparse representation.”
Scene classification based on single-layer SAE and SVM (2015, background): “Recently, increasing attention has been drawn to the study of single-layer SAE (Coates et al., 2011; Deng et al., 2013).”
Introducing shared-hidden-layer autoencoders for transfer learning and their application in acoustic emotion recognition (2014, background): “Recently, [10] uses sparse autoencoders to transfer useful knowledge from other corpora to the target one using the labels of the target set.”
Deep Learning Approach Combining Sparse Autoencoder With SVM for Network Intrusion Detection (2018, methodology): “The wide application of STL extends particularly to image identification [32], [33], SVM for classification tasks, and distinguishing different types of intrusions because combining robust classifiers, such as SVM, with SAE leads to enhanced performance in intrusion detection.”
Evaluating Lottery Tickets Under Distributional Shifts (2019, background): “There is a large body of work on transfer learning for neural networks (Deng et al., 2013; Yosinski et al., 2014; Liu et al., 2017; Zoph et al., 2018; Kornblith et al., 2019).”
Learning Classifiers from Synthetic Data Using a Multichannel Autoencoder (2015, background): “Recently autoencoder with its different variants [9], [18] also exhibit the success in learning and transferring sharing knowledge among data source from different domains [3], [5], [11], thus beneﬁt other machine learning tasks.”
Exploiting time-frequency patterns with LSTM-RNNs for low-bitrate audio restoration (2019, methodology): “The proposed method fundamentally extends the basic autoencoder structure [7, 8, 10, 18], which directly learns a nonlinear spectral mapping from compressed audio spectrum to uncompressed audio spectrum, allowing to reconstruct high-quality audio from a low-quality version.”
An Introductory Review of Deep Learning for Prediction Models With Big Data (2020, methodology): “We would like to highlight that there are many extensions of these models, e.g., sparse autoencoder, denoising autoencoder or variational autoencoder (Vincent et al., 2010; Deng et al., 2013; Pu et al., 2016).”
Machine Learning and Deep Learning Methods for Intrusion Detection Systems: A Survey (2019, background): “Many famous autoencoder variants exist, such as denoising autoencoders [14,15] and sparse autoencoders [16].”
CityTransfer (2018, methodology): “3) To ensure the features of each location grid preserve high-quality semantics, we use AutoEncoder [4,10,29] to construct original features of each location grid, which makes the new features become more robust and informative and further help improve the quality of chain store recommendation.”
Learning Salient Features for Speech Emotion Recognition Using Convolutional Neural Networks (2014, methodology): “In [6], a common emotion-specific mapping rule is learnt from a long feature vector.”