Abstract
With the increasing production of hyper-realistic altered images, the need for effective deepfake detection technology has become critical. These altered images pose significant threats to security, privacy, and the spread of misinformation, complicating the distinction between authentic and manipulated content. This challenge has far-reaching implications, from social media and politics to personal relationships. This study focuses on detecting deepfake human face images using a balanced dataset of 140,000 images, comprising 70,000 real faces sourced from Nvidia’s Flickr dataset and 70,000 fake faces generated by StyleGAN. In this research, we compare the performance of EfficientNetB4 and VGG19 models, to identify subtle manipulations in high-quality deepfake images. Our findings demonstrate that the EfficientNetB4 model achieves a test accuracy of 98.54%, while the VGG19 model reaches 99.11% in the base model, highlighting the effectiveness of these models in advancing deepfake detection technology.
Keywords:
Deepfake Detection, GANs, Convolutional Neural Network, EfficientNetB4, VGG19
Abstract
The Federal Reserve’s interest rate decisions significantly influence macroeconomic stability, affecting the financial markets, investments, and household finances. Traditionally, interest rate decisions have been made by analyzing certain economic indicators like the unemployment rate, inflation rate, and GDP. However, in recent years the traditional models have not been considered sufficient and enhanced to capture the fed decision in the interest rate policies. So, to increase the accuracy and interpretability of future economic conditions specifically with interest rates, the combination of sentiment analysis of unstructured textual data such as Federal Reserve economic reports and members’ speeches and meeting minutes, and traditional numerical data analysis would help to enhance the accuracy of the predictions. This research proposes a hybrid machine learning framework integrating both textual and numerical data to predict Federal Reserve interest rate decisions. Separate pipelines will preprocess and analyze the numerical and textual data using machine learning models (such as Random Forest, Gradient Boosting, and BERT). The outputs from these two pipelines will be combined through ensemble modeling techniques to improve predictive accuracy. This innovative approach not only enhances the understanding of interest rate decisions but also provides valuable insights for policymakers, investors, and the public, ensuring a comprehensive analysis of economic trends.
Keywords:
Federal Reserve, Interest Rate Prediction, Machine Learning, Sentiment analysis, Economic Indicators, Fed Policy, Economic News and Report Analysis, Ensemble Techniques Deepfake Detection, GANs, Convolutional Neural Network, EfficientNetB4, VGG19