This review provides a comprehensive analysis of machine learning techniques for human emotion recognition (AHER), focusing on the latest advancements and challenges in the field. Emotion recognition is a critical area in computer science, with applications in marketing, human-robot interaction, electronic games, e-learning, and more. AHER systems use various modalities such as facial expressions, speech, text, and physiological signals (e.g., electroencephalograph, blood volume pulse, electrocardiogram) to recognize and classify emotions. These signals can be used individually or in combination (multi-modal) to improve recognition accuracy. The review discusses the evolution of emotion research, highlighting the contributions of Charles Darwin, who proposed that emotions are universal and expressed similarly across species. Emotion models, such as the circumplex model, Lovheim’s emotional cube, and Plutchik’s emotion wheel, are used to categorize and understand emotions. These models are essential for developing effective emotion recognition systems. The review also covers various machine learning algorithms, including supervised and unsupervised learning techniques, and deep learning methods such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), and long short-term memory (LSTM) networks. These algorithms are used to classify and predict emotions based on different modalities. Supervised learning methods like support vector machines (SVM), random forests, and gradient boosted machines are widely used for emotion classification. Unsupervised learning techniques such as k-means clustering and self-organizing maps are also discussed. Deep learning techniques, including deep belief networks (DBNs) and deep neural networks (DNNs), are highlighted for their ability to handle complex patterns in emotional data. Generative models such as autoencoders, variational autoencoders (VAEs), and generative adversarial networks (GANs) are also discussed for their potential in emotion recognition. These models can generate synthetic data to improve training and enhance the accuracy of emotion recognition systems. The review emphasizes the importance of using both lab-based and real-world experiments to develop generic and robust emotion recognition models. It also discusses the challenges of emotion recognition, including the need for accurate and consistent emotional definitions, the complexity of human emotional expressions, and the impact of environmental factors on physiological responses. The review concludes with recommendations for future research directions, including the development of more efficient and accurate emotion recognition systems.This review provides a comprehensive analysis of machine learning techniques for human emotion recognition (AHER), focusing on the latest advancements and challenges in the field. Emotion recognition is a critical area in computer science, with applications in marketing, human-robot interaction, electronic games, e-learning, and more. AHER systems use various modalities such as facial expressions, speech, text, and physiological signals (e.g., electroencephalograph, blood volume pulse, electrocardiogram) to recognize and classify emotions. These signals can be used individually or in combination (multi-modal) to improve recognition accuracy. The review discusses the evolution of emotion research, highlighting the contributions of Charles Darwin, who proposed that emotions are universal and expressed similarly across species. Emotion models, such as the circumplex model, Lovheim’s emotional cube, and Plutchik’s emotion wheel, are used to categorize and understand emotions. These models are essential for developing effective emotion recognition systems. The review also covers various machine learning algorithms, including supervised and unsupervised learning techniques, and deep learning methods such as convolutional neural networks (CNNs), recurrent neural networks (RNNs), and long short-term memory (LSTM) networks. These algorithms are used to classify and predict emotions based on different modalities. Supervised learning methods like support vector machines (SVM), random forests, and gradient boosted machines are widely used for emotion classification. Unsupervised learning techniques such as k-means clustering and self-organizing maps are also discussed. Deep learning techniques, including deep belief networks (DBNs) and deep neural networks (DNNs), are highlighted for their ability to handle complex patterns in emotional data. Generative models such as autoencoders, variational autoencoders (VAEs), and generative adversarial networks (GANs) are also discussed for their potential in emotion recognition. These models can generate synthetic data to improve training and enhance the accuracy of emotion recognition systems. The review emphasizes the importance of using both lab-based and real-world experiments to develop generic and robust emotion recognition models. It also discusses the challenges of emotion recognition, including the need for accurate and consistent emotional definitions, the complexity of human emotional expressions, and the impact of environmental factors on physiological responses. The review concludes with recommendations for future research directions, including the development of more efficient and accurate emotion recognition systems.