Dataset Information

Deepfake tweets classification using stacked Bi-LSTM and words embedding.

ABSTRACT: The spread of altered media in the form of fake videos, audios, and images, has been largely increased over the past few years. Advanced digital manipulation tools and techniques make it easier to generate fake content and post it on social media. In addition, tweets with deep fake content make their way to social platforms. The polarity of such tweets is significant to determine the sentiment of people about deep fakes. This paper presents a deep learning model to predict the polarity of deep fake tweets. For this purpose, a stacked bi-directional long short-term memory (SBi-LSTM) network is proposed to classify the sentiment of deep fake tweets. Several well-known machine learning classifiers are investigated as well such as support vector machine, logistic regression, Gaussian Naive Bayes, extra tree classifier, and AdaBoost classifier. These classifiers are utilized with term frequency-inverse document frequency and a bag of words feature extraction approaches. Besides, the performance of deep learning models is analyzed including long short-term memory network, gated recurrent unit, bi-direction LSTM, and convolutional neural network+LSTM. Experimental results indicate that the proposed SBi-LSTM outperforms both machine and deep learning models and achieves an accuracy of 0.92.

SUBMITTER: Rupapara V

PROVIDER: S-EPMC8576542 | biostudies-literature | 2021

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Deepfake tweets classification using stacked Bi-LSTM and words embedding.

Rupapara Vaibhav V Rustam Furqan F Amaar Aashir A Washington Patrick Bernard PB Lee Ernesto E Ashraf Imran I

PeerJ. Computer science 20211021

The spread of altered media in the form of fake videos, audios, and images, has been largely increased over the past few years. Advanced digital manipulation tools and techniques make it easier to generate fake content and post it on social media. In addition, tweets with deep fake content make their way to social platforms. The polarity of such tweets is significant to determine the sentiment of people about deep fakes. This paper presents a deep learning model to predict the polarity of deep f ...[more]

PMID: 34805502

Similar Datasets

Project description:The prediction of energy consumption in households is essential due to the reliance on electrical appliances for daily activities. Accurate assessment of energy demand is crucial for effective energy generation, preventing overloads and optimizing energy storage. Traditional techniques have limitations in accuracy and error rates, necessitating advancements in prediction techniques. To enhance prediction accuracy, a proposed smart city system utilizes the Household Energy Consumption dataset, employing deep learning algorithms. In the beginning, data pre-processing addresses missing values and performs feature scaling for normalizing independent variables. Followed by that, Modified Deep CNN-Bi-LSTM (Convolutional Neural Network and Bi-directional Long Short Term Memory) with attention mechanism is utilized for regression which extracts temporal and spatial complex features. Deep CNN extracts features impacting energy consumption whereas Bi-LSTM with attention layer finds suitability for regression as it is capable of modelling irregular trends in the time-series components, where the attention mechanism is implemented to enhance the decoder's ability to selectively focus on the most relevant segments of the input sequence. This is achieved through a weighted integration of all encoded input trajectories, allowing the model to dynamically emphasize the vectors that carry the highest significance for accurate predictions. Based on regression outcomes from analysis taken in hourly, daily and monthly time intervals, enhanced prediction accuracy is estimated through evaluation metrics such as MSE (Mean Square Error), MAPE (Mean Absolute Percentage Error) and RMSE (Root Mean Square Error) which determines the efficacy of the system, where Specifically, the proposed model achieves MSE of 0.123, MAE of 0.22, and MAPE of 324.12. Furthermore, this model demonstrates a training time of 692.12 s and a prediction time of just 1.87 s. Therefore, present research highlights the critical need for accurate energy consumption prediction in households, driven by the increasing reliance on electrical appliances in daily life.

Project description:BackgroundWith the increasing variety of drugs, the incidence of adverse drug events (ADEs) is increasing year by year. Massive numbers of ADEs are recorded in electronic medical records and adverse drug reaction (ADR) reports, which are important sources of potential ADR information. Meanwhile, it is essential to make latent ADR information automatically available for better postmarketing drug safety reevaluation and pharmacovigilance.ObjectiveThis study describes how to identify ADR-related information from Chinese ADE reports.MethodsOur study established an efficient automated tool, named BBC-Radical. BBC-Radical is a model that consists of 3 components: Bidirectional Encoder Representations from Transformers (BERT), bidirectional long short-term memory (bi-LSTM), and conditional random field (CRF). The model identifies ADR-related information from Chinese ADR reports. Token features and radical features of Chinese characters were used to represent the common meaning of a group of words. BERT and Bi-LSTM-CRF were novel models that combined these features to conduct named entity recognition (NER) tasks in the free-text section of 24,890 ADR reports from the Jiangsu Province Adverse Drug Reaction Monitoring Center from 2010 to 2016. Moreover, the man-machine comparison experiment on the ADE records from Drum Tower Hospital was designed to compare the NER performance between the BBC-Radical model and a manual method.ResultsThe NER model achieved relatively high performance, with a precision of 96.4%, recall of 96.0%, and F1 score of 96.2%. This indicates that the performance of the BBC-Radical model (precision 87.2%, recall 85.7%, and F1 score 86.4%) is much better than that of the manual method (precision 86.1%, recall 73.8%, and F1 score 79.5%) in the recognition task of each kind of entity.ConclusionsThe proposed model was competitive in extracting ADR-related information from ADE reports, and the results suggest that the application of our method to extract ADR-related information is of great significance in improving the quality of ADR reports and postmarketing drug safety evaluation.

Project description:BackgroundDeepfakes are fake images or videos generated by deep learning algorithms. Ongoing progress in deep learning techniques like auto-encoders and generative adversarial networks (GANs) is approaching a level that makes deepfake detection ideally impossible. A deepfake is created by swapping videos, images, or audio with the target, consequently raising digital media threats over the internet. Much work has been done to detect deepfake videos through feature detection using a convolutional neural network (CNN), recurrent neural network (RNN), and spatiotemporal CNN. However, these techniques are not effective in the future due to continuous improvements in GANs. Style GANs can create fake videos with high accuracy that cannot be easily detected. Hence, deepfake prevention is the need of the hour rather than just mere detection.MethodsRecently, blockchain-based ownership methods, image tags, and watermarks in video frames have been used to prevent deepfake. However, this process is not fully functional. An image frame could be faked by copying watermarks and reusing them to create a deepfake. In this research, an enhanced modified version of the steganography technique RivaGAN is used to address the issue. The proposed approach encodes watermarks into features of the video frames by training an "attention model" with the ReLU activation function to achieve a fast learning rate.ResultsThe proposed attention-generating approach has been validated with multiple activation functions and learning rates. It achieved 99.7% accuracy in embedding watermarks into the frames of the video. After generating the attention model, the generative adversarial network has trained using DeepFaceLab 2.0 and has tested the prevention of deepfake attacks using watermark embedded videos comprising 8,074 frames from different benchmark datasets. The proposed approach has acquired a 100% success rate in preventing deepfake attacks. Our code is available at https://github.com/shahidmuneer/deepfakes-watermarking-technique.

Dataset Information

Deepfake tweets classification using stacked Bi-LSTM and words embedding.

Publications

Deepfake tweets classification using stacked Bi-LSTM and words embedding.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets