Multimodal sentiment analysis / Soujanya Poria, Amir Hussain, Erik Cambria.

Intro; Preface; Contents; 1 Introduction and Motivation; 1.1 Research Challenges in Text-Based Sentiment Analysis; 1.2 Research Challenges in Multimodal Sentiment Analysis; 1.3 Overview of the Proposed Framework; 1.3.1 Text-Based Sentiment Analysis; 1.3.2 Multimodal Sentiment Analysis; 1.4 Contributions of This Book; 1.4.1 Text-Based Sentiment Analysis; 1.4.2 Multimodal Sentiment Analysis; 1.5 Book Organisation; 2 Background; 2.1 Affective Computing; 2.2 Sentiment Analysis; 2.2.1 Opinion Holder; 2.2.2 Aspects; 2.2.3 Subjectivity; 2.3 Pattern Recognition; 2.3.1 Features; 2.3.2 Pattern

2.3.3 Class2.4 Feature Selection; 2.4.1 Principal Component Analysis; 2.5 Model Evaluation Techniques; 2.5.1 Evaluating Regression Quality; 2.5.1.1 Mean Squared Error Validation Techniques; 2.5.2 Evaluating Classification Techniques; 2.5.2.1 Precision; 2.5.2.2 Recall; 2.5.2.3 F-Score; 2.5.2.4 Accuracy; 2.6 Model Validation Techniques; 2.6.1 Cross Validation; 2.6.2 Bootstrapping; 2.7 Classification Techniques; 2.7.1 Support Vector Machine; 2.7.1.1 Soft-Margin Extension; 2.7.1.2 Non-linear Decision Boundary; 2.7.1.3 Formulation as a Lagrangian Optimization; 2.7.1.4 Kernel Trick

2.7.2 Extreme Learning Machine2.7.3 Deep Neural Networks; 2.8 Feature-Based Text Representation; 2.8.1 Types of Feature-Based Text Representation; 2.8.1.1 One-Hot Vectors; 2.8.1.2 Distributional Vectors; 2.8.1.3 Word Embeddings; 2.8.2 Word2Vec; 2.8.2.1 Negative Sampling; 2.8.2.2 Classification Problem; 2.9 Conclusion; 3 Literature Survey and Datasets; 3.1 Introduction; 3.2 Available Datasets; 3.2.1 Datasets for Multimodal Sentiment Analysis; 3.2.1.1 YouTube Dataset; 3.2.1.2 MOUD Dataset; 3.2.1.3 ICT-MMMO Database; 3.2.2 Datasets for Multimodal Emotion Recognition; 3.2.2.1 HUMAINE Database

3.2.2.2 The Belfast Database3.2.2.3 The SEMAINE Database; 3.2.2.4 Interactive Emotional Dyadic Motion Capture Database (IEMOCAP); 3.2.2.5 The eNTERFACE Database; 3.2.3 Affective Detection from Textual Modality; 3.2.3.1 Single- vs. Cross-Domain; 3.2.3.2 Use of Linguistic Patterns; 3.2.3.3 Bag of Words Versus Bag of Concepts; 3.2.3.4 Contextual Subjectivity; 3.2.3.5 New Era of NLP: Emergence of Deep Learning; 3.3 Visual, Audio Features for Affect Recognition; 3.3.1 Visual Modality; 3.3.1.1 Facial Action Coding System; 3.3.1.2 Main Facial Expression Recognition Techniques

3.3.1.3 Extracting Temporal Features from Videos3.3.1.4 Body Gestures; 3.3.1.5 New Era: Deep Learning to Extract Visual Features; 3.3.2 Audio Modality; 3.3.2.1 Local Features vs. Global Features; 3.3.2.2 Speaker-Independent Applications; 3.3.2.3 Audio Features Extraction Using Deep Networks; 3.4 Multimodal Affect Recognition; 3.4.1 Information Fusion Techniques; 3.4.1.1 Decision-Level or Late Fusion alam2014predicting,cai2015convolutional, yamasaki2015prediction, glodek2013kalman, dobrivsek2013towards; 3.4.1.2 Hybrid Multimodal Fusion wollmer2013YouTube,mansoorizadeh2010multimodal