A Novel Policy for Pre-trained Deep Reinforcement Learning for Speech Emotion Recognition (2021-01-04T00:00:00.000000Z)

TL;DR

A novel policy, the “Zeta policy” tailored for SER is introduced and pre- training in deep RL is introduced to achieve a faster learning rate and support that pre-training can reduce training time and is robust to a cross-corpus scenario.

Abstract

Deep Reinforcement Learning (deep RL) has gained tremendous success in gaming but it has rarely been explored for Speech Emotion Recognition (SER). In the RL literature, policy used by the RL agent plays a major role in action selection, however, there is no RL policy tailored for SER. Also, an extended learning period is a general challenge for deep RL, which can impact the speed of learning for SER. In this paper, we introduce a novel policy, the “Zeta policy” tailored for SER and introduce pre-training in deep RL to achieve a faster learning rate. Pre-training with a cross dataset was also studied to discover the feasibility of pre-training the RL agent with a similar dataset in a scenario where real environmental data is not available. We use “IEMOCAP” and “SAVEE” datasets for the evaluation with the problem of recognising four emotions, namely happy, sad, angry, and neutral. The experimental results show that the proposed policy performs better than existing policies. Results also support that pre-training can reduce training time and is robust to a cross-corpus scenario.

Authors

Björn Schuller

10 papers

R. Rana

2 papers

Thejan Rajapakshe

1 papers

TL;DR

Abstract

Authors

References56 items

Semi-supervised Training for Sequence-to-Sequence Speech Recognition Using Reinforcement Learning

Speech emotion recognition with deep convolutional neural networks

Deep Representation Learning in Speech Processing: Challenges, Recent Advances, and Future Trends

Grandmaster level in StarCraft II using multi-agent reinforcement learning

Pre-training in Deep Reinforcement Learning for Automatic Speech Recognition

An Emotion-Based Approach to Reinforcement Learning Reward Design

Speech Emotion Recognition Using Deep Neural Network Considering Verbal and Nonverbal Speech Sounds

Direct Modelling of Speech Emotion from Raw Speech

Automated Screening for Distress: A Perspective for the Future

A Theoretical Analysis of Deep Q-Learning

Adversarial Machine Learning And Speech Emotion Recognition: Utilizing Generative Adversarial Networks For Robustness

Reinforcement Learning Based Speech Enhancement for Robust Speech Recognition

A Deep Learning-based Stress Detection Algorithm with Speech Signal

Model-Based Regularization for Deep Reinforcement Learning with Transcoder Networks

Multi-Modal Emotion recognition on IEMOCAP Dataset using Deep Learning

EmoRL: Continuous Acoustic Emotion Classification Using Deep Reinforcement Learning

CNN+LSTM Architecture for Speech Emotion Recognition with Data Augmentation

Cross Corpus Speech Emotion Classification- An Effective Transfer Learning Technique

Reinforcement Learning To Adapt Speech Enhancement to Instantaneous Input Signal Quality

Pre-training Neural Networks with Human Demonstrations for Deep Reinforcement Learning

StarCraft II: A New Challenge for Reinforcement Learning

A breakthrough in Speech emotion recognition using Deep Retinal Convolution Neural Networks

The Atari Grand Challenge Dataset

Boltzmann Exploration Done Right

Deep Q-learning From Demonstrations

Mastering the game of Go with deep neural networks and tree search

Continuous control with deep reinforcement learning

Graph-based semi-supervised acoustic modeling in DNN-based speech recognition

Learning Salient Features for Speech Emotion Recognition Using Convolutional Neural Networks

Speech emotion recognition using deep neural network and extreme learning machine

Deep neural network features and semi-supervised training for low resource speech recognition

Survey on speech emotion recognition: Features, classification schemes, and databases

Roles of Pre-Training and Fine-Tuning in Context-Dependent DBN-HMMs for Real-World Speech Recognition

Incremental Learning by Heterogeneous Bagging Ensemble

Reinforcement Learning of Listener Response for Mood Classification of Audio

IEMOCAP: interactive emotional dyadic motion capture database

Audio-visual feature selection and reduction for emotion classification

Emotional speech recognition: Resources, features, and methods

Reinforcement Learning for Spoken Dialogue Systems: Comparing Strengths and Weaknesses for Practical Deployment

Reinforcement Learning as Classification: Leveraging Modern Classifiers

Hidden Markov model-based speech emotion recognition

Technical Update: Least-Squares Temporal Difference Learning

Combining acoustic and language information for emotion recognition

Learn++: an incremental learning algorithm for supervised neural networks

Reinforcement Learning for Spoken Dialogue Systems

Explorations in efficient reinforcement learning

Learning from delayed rewards

Nonlinear analysis and classification of speech under stressed conditions

Reinforcement learning with dynamic boltzmann softmax updates

Learning from Demonstration ( Programming by Demonstration ) ∗

Comparison of Parametric Representation for Monosyllabic Word Recognition in Continuously Spoken Sentences

Cross dataset pre-training by IEMOCAP (IEM ) dataset and SAVEE (SAV ) dataset neural networks and tree search

librosa: Audio and Music Signal Analysis in Python

TensorFlow: Large-Scale Machine Learning on Heterogeneous Systems

François Chollet and others

The NIPS workshop on Deep Learning and Unsupervised Feature Learning

Field of Study

Journal Information

Name

Venue Information

Name

Type

URL

Alternate Names