Emotion Dataset Research Articles

A Study of Potential Applications of Student Emotion Recognition in Primary and Secondary Classrooms

Emotion recognition is critical to understanding students’ emotional states. However, problems such as crowded classroom environments, changing light, and occlusion often affect the accuracy of recognition. This study proposes an emotion recognition algorithm specifically for classroom environments. Firstly, the study adds the self-made MCC module and the Wise-IoU loss function to make object detection in the YOLOv8 model more accurate and efficient. Compared with the native YOL0v8x, it reduces the parameters by 16% and accelerates the inference speed by 20%. Secondly, in order to address the intricacies of the classroom setting and the specific requirements of the emotion recognition task, a multi-channel emotion recognition network (MultiEmoNet) has been developed. This network fuses skeletal, environmental, and facial information, and introduces a central loss function and an attention module AAM to enhance the feature extraction capability. The experimental results show that MultiEmoNet achieves a classification accuracy of 91.4% on a homemade classroom student emotion dataset, which is a 10% improvement over the single-channel classification algorithm. In addition, this study also demonstrates the dynamic changes in students’ emotions in the classroom through visual analysis, which helps teachers grasp students’ emotional states in real time. This paper validates the potential of multi-channel information-fusion deep learning techniques for classroom teaching analysis and provides new ideas and tools for future improvements to emotion recognition techniques.

LCANet: a model for analysis of students real-time sentiment by integrating attention mechanism and joint loss function

By recognizing students’ facial expressions in actual classroom situations, the students’ emotional states can be quickly uncovered, which can help teachers grasp the students’ learning rate, which allows teachers to adjust their teaching strategies and methods, thus improving the quality and effectiveness of classroom teaching. However, most previous facial expression recognition methods have problems such as missing key facial features and imbalanced class distributions in the dateset, resulting in low recognition accuracy. To address these challenges, this paper proposes LCANet, a model founded on a fused attention mechanism and a joint loss function, which allows the recognition of students’ emotions in real classroom scenarios. The model uses ConvNeXt V2 as the backbone network to optimize the global feature extraction capability of the model, and at the same time, it enables the model to pay closer attention to the key regions in facial expressions. We incorporate an improved Channel Spatial Attention (CSA) module as a way to extract more local feature information. Furthermore, to mitigate the class distribution imbalance problem in the facial expression dataset, we introduce a joint loss function. The experimental results show that our LCANet model has good recognition rates on both the public emotion datasets FERPlus, RAF-DB and AffectNet, with accuracies of 91.43%, 90.03% and 64.43%, respectively, with good robustness and generalizability. Additionally, we conducted experiments using the model in real classroom scenarios, detecting and accurately predicting students’ classroom emotions in real time, which provides an important reference for improving teaching in smart teaching scenarios.

Speech Emotion Recognition Using Transfer Learning: Integration of Advanced Speaker Embeddings and Image Recognition Models

Automatic Speech Emotion Recognition (SER) plays a vital role in making human–computer interactions more natural and effective. A significant challenge in SER development is the limited availability of diverse emotional speech datasets, which hinders the application of advanced deep learning models. Transfer learning is a machine learning technique that helps address this issue by utilizing knowledge from pre-trained models to improve performance on a new task in a target domain, even with limited data. This study investigates the use of transfer learning from various pre-trained networks, including speaker embedding models such as d-vector, x-vector, and r-vector, and image classification models like AlexNet, GoogLeNet, SqueezeNet, ResNet-18, and ResNet-50. We also propose enhanced versions of the x-vector and r-vector models incorporating Multi-Head Attention Pooling and Angular Margin Softmax, alongside other architectural improvements. Additionally, reverberation from the Room Impulse Response datasets was added to the speech utterances to diversify and augment the available data. Notably, the enhanced r-vector model achieved classification accuracies of 74.05% Unweighted Accuracy (UA) and 73.68% Weighted Accuracy (WA) on the IEMOCAP dataset, and 80.25% UA and 79.81% WA on the CREMA-D dataset, outperforming the existing state-of-the-art methods. This study shows that using cross-domain transfer learning is beneficial for low-resource emotion recognition. The enhanced models developed in other domains (for non-emotional tasks) can further improve the accuracy of SER.

Multimodal emotion recognition: A comprehensive review, trends, and challenges

AbstractAutomatic emotion recognition is a burgeoning field of research and has its roots in psychology and cognitive science. This article comprehensively reviews multimodal emotion recognition, covering various aspects such as emotion theories, discrete and dimensional models, emotional response systems, datasets, and current trends. This article reviewed 179 multimodal emotion recognition literature papers from 2017 to 2023 to reflect on the current trends in multimodal affective computing. This article covers various modalities used in emotion recognition based on the emotional response system under four categories: subjective experience comprising text and self‐report; peripheral physiology comprising electrodermal, cardiovascular, facial muscle, and respiration activity; central physiology comprising EEG, neuroimaging, and EOG; behavior comprising facial, vocal, whole‐body behavior, and observer ratings. This review summarizes the measures and behavior of each modality under various emotional states. This article provides an extensive list of multimodal datasets and their unique characteristics. The recent advances in multimodal emotion recognition are grouped based on the research focus areas such as emotion elicitation strategy, data collection and handling, the impact of culture and modality on multimodal emotion recognition systems, feature extraction, feature selection, alignment of signals across the modalities, and fusion strategies. The recent multimodal fusion strategies are detailed in this article, as extracting shared representations of different modalities, removing redundant features from different modalities, and learning critical features from each modality are crucial for multimodal emotion recognition. This article summarizes the strengths and weaknesses of multimodal emotion recognition based on the review outcome, along with challenges and future work in multimodal emotion recognition. This article aims to serve as a lucid introduction, covering all aspects of multimodal emotion recognition for novices.This article is categorized under: Fundamental Concepts of Data and Knowledge > Human Centricity and User Interaction Technologies > Cognitive Computing Technologies > Artificial Intelligence

Talking Face Generation With Audio-Deduced Emotional Landmarks.

The goal of talking face generation is to synthesize a sequence of face images of the specified identity, ensuring the mouth movements are synchronized with the given audio. Recently, image-based talking face generation has emerged as a popular approach. It could generate talking face images synchronized with the audio merely depending on a facial image of arbitrary identity and an audio clip. Despite the accessible input, it forgoes the exploitation of the audio emotion, inducing the generated faces to suffer from emotion unsynchronization, mouth inaccuracy, and image quality deficiency. In this article, we build a bistage audio emotion-aware talking face generation (AMIGO) framework, to generate high-quality talking face videos with cross-modally synced emotion. Specifically, we propose a sequence-to-sequence (seq2seq) cross-modal emotional landmark generation network to generate vivid landmarks, whose lip and emotion are both synchronized with input audio. Meantime, we utilize a coordinated visual emotion representation to improve the extraction of the audio one. In stage two, a feature-adaptive visual translation network is designed to translate the synthesized landmarks into facial images. Concretely, we proposed a feature-adaptive transformation module to fuse the high-level representations of landmarks and images, resulting in significant improvement in image quality. We perform extensive experiments on the multi-view emotional audio-visual dataset (MEAD) and crowd-sourced emotional multimodal actors dataset (CREMA-D) benchmark datasets, demonstrating that our model outperforms state-of-the-art benchmarks.

EAV: EEG-Audio-Video Dataset for Emotion Recognition in Conversational Contexts

Understanding emotional states is pivotal for the development of next-generation human-machine interfaces. Human behaviors in social interactions have resulted in psycho-physiological processes influenced by perceptual inputs. Therefore, efforts to comprehend brain functions and human behavior could potentially catalyze the development of AI models with human-like attributes. In this study, we introduce a multimodal emotion dataset comprising data from 30-channel electroencephalography (EEG), audio, and video recordings from 42 participants. Each participant engaged in a cue-based conversation scenario, eliciting five distinct emotions: neutral, anger, happiness, sadness, and calmness. Throughout the experiment, each participant contributed 200 interactions, which encompassed both listening and speaking. This resulted in a cumulative total of 8,400 interactions across all participants. We evaluated the baseline performance of emotion recognition for each modality using established deep neural network (DNN) methods. The Emotion in EEG-Audio-Visual (EAV) dataset represents the first public dataset to incorporate three primary modalities for emotion recognition within a conversational context. We anticipate that this dataset will make significant contributions to the modeling of the human emotional process, encompassing both fundamental neuroscience and machine learning viewpoints.

Analysing Psychological Sentiment Prediction Across Modalities: Harnessing Emotion Datasets within Natural Language Processing (NLP)

Studying human emotions and feelings is a crucial element in the field of psychology, having significant implications such as evaluating mental health and improving human-computer interactions. Recently, there has been a rise in interest in examining how psychological sentiment can be predicted through various mediums, including text, audio, video, and physiological signals. By utilizing advancements in Natural Language Processing (NLP) and analysing multimodal data, this research delves into incorporating emotion datasets into NLP frameworks to improve psychological sentiment prediction. This presents present an applicability of some of the NLP techniques to predicts the Psychological Sentiment such as Deep Generating adversarial networks (D-GANs), Long short-term memory (LSTM) and gated recurrent unit (GRU). The sentimental analysis performed by the considered algorithms are implemented in python and the parameters which are considered for the results evaluation are model loss, confusion matrix, accuracy, precision, recall and f1-score. Our goal with this analysis is to offer a deeper understanding of the existing methodologies and future scope of growth in psychological sentiment prediction of research in the field of NLP.

Brain Emotion Perception Inspired EEG Emotion Recognition With Deep Reinforcement Learning.

Inspired by the well-known Papez circuit theory and neuroscience knowledge of reinforcement learning, a double dueling deep Q network (DQN) is built incorporating the electroencephalogram (EEG) signals of the frontal lobe as prior information, which is named frontal lobe double dueling DQN (FLD3QN). The framework of FLD3QN is constructed in accord with the brain emotion mechanism which takes the frontal lobe and the thalamus as the core, in which the part of the Papez circuit is simulated by the bifrontal lobe residual convolution neural network (BiFRCNN). Moreover, a step penalty factor is designed to constrain the number of mistakes of the agent. The ablation studies results on the public EEG emotion dataset DEAP verified the important roles of the frontal lobe and the Papez circuit in modeling the procedure of learning rewards during the perception of emotions, with a great increase in the average accuracies by 25.24% and 23.31% in valence and arousal dimensions.

Advancing emotion recognition in facial expressions through PCA, RFE, and MLP Integration

Emotion recognition through facial expressions is vital for enhancing human-computer interaction, making systems more intuitive and responsive to user needs. This study introduces an innovative approach to emotion detection, leveraging Principal Component Analysis (PCA) and Recursive Feature Elimination (RFE) for feature extraction and optimization. The methodology focuses on refining facial feature representations to improve classification accuracy, which is critical for accurately detecting emotions like happiness, sadness, fear, and surprise. The approach was applied to two widely recognized facial emotion datasets: the Cohn-Kanade (CK) and the Japanese Female Facial Expression (JAFFE) datasets. By integrating PCA and RFE, the model efficiently selects the most relevant features, enhancing the overall performance of emotion recognition. A comparative analysis with existing deep learning models highlights the advantages of the proposed method. The effectiveness of this approach is further supported by the results, where the model achieves an accuracy of 98.49% on the CK dataset and 96.03% on the JAFFE dataset. These results demonstrate a significant improvement over recent methods, indicating the model's potential for real-world applications.

EEG-fNIRS-Based Emotion Recognition Using Graph Convolution and Capsule Attention Network.

Electroencephalogram (EEG) and functional near-infrared spectroscopy (fNIRS) can objectively reflect a person's emotional state and have been widely studied in emotion recognition. However, the effective feature fusion and discriminative feature learning from EEG-fNIRS data is challenging. In order to improve the accuracy of emotion recognition, a graph convolution and capsule attention network model (GCN-CA-CapsNet) is proposed. Firstly, EEG-fNIRS signals are collected from 50 subjects induced by emotional video clips. And then, the features of the EEG and fNIRS are extracted; the EEG-fNIRS features are fused to generate higher-quality primary capsules by graph convolution with the Pearson correlation adjacency matrix. Finally, the capsule attention module is introduced to assign different weights to the primary capsules, and higher-quality primary capsules are selected to generate better classification capsules in the dynamic routing mechanism. We validate the efficacy of the proposed method on our emotional EEG-fNIRS dataset with an ablation study. Extensive experiments demonstrate that the proposed GCN-CA-CapsNet method achieves a more satisfactory performance against the state-of-the-art methods, and the average accuracy can increase by 3-11%.

Enhancing speech emotion detection with Windowed Long-Term Average Spectrum and Logistic-Rectified Linear Unit

Speech emotion detection can help monitor people's mental health by analyzing their speech patterns and recognizing emotional indicators. The raw data is initially augmented using various techniques such as shifting, pitching, noise injection, and stretching. Some feature extraction methods may only produce a limited set of features, potentially missing out on important information present in the data. This can result in reduced accuracy and the inability to capture complex patterns and variations in speech signals. The planned work consists of two phases: attribute extraction and classification. In the first phase, the WLTAS (Windowed Long-Term Average Spectrum) feature extraction extracts spectral information from speech signals, providing a compact and representative feature representation. It captures the frequency content and energy distribution of speech over time, enabling the analysis of important spectral characteristics. In the second phase, the Logistic-Rectified Linear Unit (LoRLU) introduces sparsity. It enhances the discriminative power of the classifier by allowing only positive values to pass through while discarding negative values. The Gibbs Restricted Boltzmann Machine (GRBM) is a generative probabilistic model that studies a joint probability delivery of the input data. It is particularly effective in capturing dependencies and patterns in high-dimensional data. The performance analysis of WLTAS-L-GRBM (Windowed Long-Term Average Spectrum - Logistic-Rectified Linear Unit - Gibbs Restricted Boltzmann Machine) has been conducted using evaluation metrics such as accuracy (99%), precision (0.97), recall (1), F1 score (0.98), MSE (0.01), and AUC-ROC (0.97) and also evaluated against various speech emotion datasets.

Differences in Fear and Negativity Levels Between Formal and Informal Health-Related Websites: Analysis of Sentiments and Emotions.

Searching for web-based health-related information is frequently performed by the public and may affect public behavior regarding health decision-making. Particularly, it may result in anxiety, erroneous, and harmful self-diagnosis. Most searched health-related topics are cancer, cardiovascular diseases, and infectious diseases. A health-related web-based search may result in either formal or informal medical website, both of which may evoke feelings of fear and negativity. Our study aimed to assess whether there is a difference in fear and negativity levels between information appearing on formal and informal health-related websites. A web search was performed to retrieve the contents of websites containing symptoms of selected diseases, using selected common symptoms. Retrieved websites were classified into formal and informal websites. Fear and negativity of each content were evaluated using 3 transformer models. A fourth transformer model was fine-tuned using an existing emotion data set obtained from a web-based health community. For formal and informal websites, fear and negativity levels were aggregated. t tests were conducted to evaluate the differences in fear and negativity levels between formal and informal websites. In this study, unique websites (N=1448) were collected, of which 534 were considered formal and 914 were considered informal. There were 1820 result pages from formal websites and 1494 result pages from informal websites. According to our findings, fear levels were statistically higher (t2753=3.331; P<.001) on formal websites (mean 0.388, SD 0.177) than on informal websites (mean 0.366, SD 0.168). The results also show that the level of negativity was statistically higher (t2753=2.726; P=.006) on formal websites (mean 0.657, SD 0.211) than on informal websites (mean 0.636, SD 0.201). Positive texts may increase the credibility of formal health websites and increase their usage by the general public and the public's compliance to the recommendations. Increasing the usage of natural language processing tools before publishing health-related information to achieve a more positive and less stressful text to be disseminated to the public is recommended.

Advancing Virtual Interviews: AI-Driven Facial Emotion Recognition for Better Recruitment

Behavior analysis involves the detailed process of identifying, modeling, and comprehending the various nuances and patterns of emotional expressions exhibited by individuals. It poses a significant challenge to accurately detect and predict facial emotions, especially in contexts like remote interviews, which have become increasingly prevalent. Notably, many participants struggle to convey their thoughts to interviewers with a happy expression and good posture, which may unfairly diminish their chances of employment, despite their qualifications. To address this challenge, artificial intelligence techniques such as image classification offer promising solutions. By leveraging AI models, behavior analysis can be applied to perceive and interpret facial reactions, thereby paving the way to anticipate future behaviors based on learned patterns to the participants. Despite existing works on facial emotion recognition (FER) using image classification, there is limited research focused on platforms like remote interviews and online courses. In this paper, our primary focus lies on emotions such as happiness, sadness, anger, surprise, eye contact, neutrality, smile, confusion, and stooped posture. We have curated our dataset, comprising a diverse range of sample interviews captured through participants' video recordings and other images documenting facial expressions and speech during interviews. Additionally, we have integrated existing datasets such as FER 2013 and the Celebrity Emotions dataset. Through our investigation, we explore a variety of AI and deep learning methodologies, including VGG19, ResNet50V2, ResNet152V2, Inception-ResNetV2, Xception, EfficientNet B0, and YOLO V8 to analyze facial patterns and predict emotions. Our results demonstrate an accuracy of 73% using the YOLO v8 model. However, we discovered that the categories of happy and smile, as well as surprised and confused, are not disjoint, leading to potential inaccuracies in classification. Furthermore, we considered stooped posture as a non-essential class since the interviews are conducted via webcam, which does not allow for the observation of posture. By removing these overlapping categories, we achieved a remarkable accuracy increase to around 76.88% using the YOLO v8 model.

EDA-Graph: Graph Signal Processing of Electrodermal Activity for Emotional States Detection.

The continuous detection of emotional states has many applications in mental health, marketing, human-computer interaction, and assistive robotics. Electrodermal activity (EDA), a signal modulated by sympathetic nervous system activity, provides continuous insight into emotional states. However, EDA possesses intricate nonstationary and nonlinear characteristics, making the extraction of emotion-relevant information challenging. We propose a novel graph signal processing (GSP) approach to model EDA signals as graphical networks, termed EDA-graph. The GSP leverages graph theory concepts to capture complex relationships in time-series data. To test the usefulness of EDA-graphs to detect emotions, we processed EDA recordings from the CASE emotion dataset using GSP by quantizing and linking values based on the Euclidean distance between the nearest neighbors. From these EDA-graphs, we computed the features of graph analysis, including total load centrality (TLC), total harmonic centrality (THC), number of cliques (GNC), diameter, and graph radius, and compared those features with features obtained using traditional EDA processing techniques. EDA-graph features encompassing TLC, THC, GNC, diameter, and radius demonstrated significant differences (p < 0.05) between five emotional states (Neutral, Amused, Bored, Relaxed, and Scared). Using machine learning models for classifying emotional states evaluated using leave-one-subject-out cross-validation, we achieved a five-class F1 score of up to 0.68.

Assessing Feature Importance in Eye-Tracking Data within Virtual Reality Using Explainable Artificial Intelligence Techniques

Our research systematically investigates the cognitive and emotional processes revealed through eye movements within the context of virtual reality (VR) environments. We assess the utility of eye-tracking data for predicting emotional states in VR, employing explainable artificial intelligence (XAI) to advance the interpretability and transparency of our findings. Utilizing the VR Eyes: Emotions dataset (VREED) alongside an extra trees classifier enhanced by SHapley Additive ExPlanations (SHAP) and local interpretable model agnostic explanations (LIME), we rigorously evaluate the importance of various eye-tracking metrics. Our results identify significant correlations between metrics such as saccades, micro-saccades, blinks, and fixations and specific emotional states. The application of SHAP and LIME elucidates these relationships, providing deeper insights into the emotional responses triggered by VR. These findings suggest that variations in eye feature patterns serve as indicators of heightened emotional arousal. Not only do these insights advance our understanding of affective computing within VR, but they also highlight the potential for developing more responsive VR systems capable of adapting to user emotions in real-time. This research contributes significantly to the fields of human-computer interaction and psychological research, showcasing how XAI can bridge the gap between complex machine-learning models and practical applications, thereby facilitating the creation of reliable, user-sensitive VR experiences. Future research may explore the integration of multiple physiological signals to enhance emotion detection and interactive dynamics in VR.

Emotion recognition in user‐generated videos with long‐range correlation‐aware network

AbstractEmotion recognition in user‐generated videos plays an essential role in affective computing. In general, visual information directly affects human emotions, so the visual modality is significant for emotion recognition. Most classic approaches mainly focus on local temporal information of videos, which potentially restricts their capacity to encode the correlation of long‐range context. To address this issue, a novel network is proposed to recognize emotions in videos. To be specific, a spatio‐temporal correlation‐aware block is designed to depict the long‐range correlations between input tokens, where the convolutional layers are used to learn the local correlations and the inter‐image cross‐attention is designed to learn the long‐range and spatio‐temporal correlations between input tokens. To generate diverse and challenging samples, a dual‐augmentation fusion layer is devised, which fuses each frame with its corresponding frame in the temporal domain. To produce rich video clips, a long‐range sampling layer is designed, which generates clips in a wide range of spatial and temporal domains. Extensive experiments are conducted on two challenging video emotion datasets, namely VideoEmotion‐8 and Ekman‐6. The experimental results demonstrate that the proposed method obtains better performance than baseline methods. Moreover, the proposed method achieves state‐of‐the‐art results on the two datasets. The source code of the proposed network is available at: https://github.com/JinChow/LRCANet.

Exploring emotional experiences and dataset construction in the era of short videos based on physiological signals

ObjectiveThis study aims to uncover the intrinsic links between emotional experiences and physiological responses as users watch short videos on social media platforms, with a particular focus on using changes in physiological signals to identify and understand different emotional states. MethodsTo achieve this objective, a simulated experiment was designed to browse short videos and induce and record physiological signals under seven typical emotional states. The recorded signals included electroencephalogram, galvanic skin response, skin temperature, and heart rate, resulting in the creation of a dataset. Machine learning algorithms were employed to classify emotions and evaluate the dataset’s effectiveness. Statistical testing methods were used to analyze signal feature changes and distributions across different emotional states, exploring trends and their statistical significance. ResultsThe study successfully constructed an emotion-physiological signal dataset. Statistical tests revealed significant changes in physiological signal characteristics across different emotional states, providing extensive data support for understanding how emotions specifically affect physiological responses. ConclusionThe research not only confirmed the practicality of the constructed dataset in emotion recognition tasks but also provided empirical evidence of how emotions influence physiological responses through detailed analysis of physiological signals. SignificanceThe findings of this study hold significant value for emotional science, psychological research, and the entertainment industry. They not only facilitate a deeper exploration of individual emotional physiological mechanisms but also provide a scientific basis for optimizing content recommendation systems, intelligent entertainment technologies, and affective-aware applications.

Optimized Multimodal Emotional Recognition Using Long Short-Term Memory

The aim of this project is to research and classification on human emotions. A new method for the recognition of speech signals has been introduced. It’s called LSTM (Long-Short Term Memory). It is a type of Recurrent neural network. RNN is used for analyzing sequential data, hence it is useful for speech signal recognition. Several Datasets were found across the internet for this project. Ex: TESS (Toronto Emotional Speech Set), RAVDESS (Ryerson Audio-Visual Database of Emotional Speech and Song), SAVEE (Surrey Audio-Visual Expressed Emotion), CREMA-D (Crowd- Sourced Emotional Multimodal Actors Dataset). The Main Dataset used in this project is TESS (Toronto Emotional Speech Set) Dataset and Mel Frequency Cepstral Coefficient (MFCC) is Used for Feature extraction.

Multi-View Multi-Label Fine-Grained Emotion Decoding From Human Brain Activity.

Decoding emotional states from human brain activity play an important role in the brain-computer interfaces. Existing emotion decoding methods still have two main limitations: one is only decoding a single emotion category from a brain activity pattern and the decoded emotion categories are coarse-grained, which is inconsistent with the complex emotional expression of humans; the other is ignoring the discrepancy of emotion expression between the left and right hemispheres of the human brain. In this article, we propose a novel multi-view multi-label hybrid model for fine-grained emotion decoding (up to 80 emotion categories) which can learn the expressive neural representations and predict multiple emotional states simultaneously. Specifically, the generative component of our hybrid model is parameterized by a multi-view variational autoencoder, in which we regard the brain activity of left and right hemispheres and their difference as three distinct views and use the product of expert mechanism in its inference network. The discriminative component of our hybrid model is implemented by a multi-label classification network with an asymmetric focal loss. For more accurate emotion decoding, we first adopt a label-aware module for emotion-specific neural representation learning and then model the dependency of emotional states by a masked self-attention mechanism. Extensive experiments on two visually evoked emotional datasets show the superiority of our method.

COMPARISON OF DEEP LEARNING TECHNIQUES FOR TEXTUAL SENTIMENT ANALYSIS

Examining the opinions of the people might provide us with valuable knowledge. Sentiment analysis is a method for analyzing textual data that helps find subjective information, such as opinions and feelings that individuals or groups have expressed. It improves our understanding of human language using deep learning and natural language techniques. Several deep learning models, including RNNs, LSTMs, GRUs, and their bidirectional variants, are compared in this work. Three publicly available datasets - the imdb_reviews, Twitter Sentiment Dataset, and Emotions dataset were used in the investigation. Accuracy performance is evaluated for six deep learning models. According to experimental studies, bidirectional structures outperform their unidirectional counterparts in most cases. Across several datasets, the bidirectional models continuously produced the best accuracy.