Predicting exchange rates is important because it affects all major markets and plays a big part in the economy. The goal of this study is to forecast future values of the Chinese Yuan (CNY) to US dollar exchange rate. One of the most crucial aspects of the economy is the rate of exchange of currencies. The currency exchange rate is required in commercial terms, such as profit and investment evaluation. The purpose of CNY rate prediction is to determine the future value of the Chinese Yuan (CNY) relative to the US dollar, which can be taken into account while making decisions and lower the chance of losing money. As a result, we require a technique that can assist in accurately assisting in business selections regarding when to execute the appropriate deals. For this research study three-year dataset is used from 25 April 2020 to 26 May 2023. For this research we used two different machine learning models and comparison between the LSTM and Xgboost models first one is Long Short-Term Memory (LSTM) and second is Extreme Gradient Boosting (XGboost). The empirical results showed that the LSTM model provided better results than XGboost. Therefore, this study suggest that the LSTM model will helpful for the government monetary policymaker, economists and other stakeholders to identify and forecast the future trend of the exchange rate and make their policies accordingly.
Deep learning models have been widely applied to chest X-ray (CXR) disease classification and diagnosis; however, challenges such as data scarcity and shortcut learning often lead to biased model behavior. This study addresses fairness-related concerns in conventional deep learning models trained on CXR data and proposes mitigating demographic disparities through image synthesis. We fine-tune a pre-trained stable diffusion model using Low-Rank Adaptation (LoRA) and a CLIP tokenizer, incorporating low-rank constraints into key attention layers while preserving the original architecture. This enables the generation of high-quality, realistic CXR images with reduced parameter complexity. Experimental results demonstrate that models trained with our synthetic data achieve improved classification performance and exhibit significantly reduced disparities across demographic groups. Furthermore, the proposed models show increased attention to disease-relevant regions and diminished reliance on spurious shortcuts. These findings highlight the potential of generative AI in enhancing fairness in medical imaging workflows, particularly when combined with efficient and adaptable fine-tuning strategies.
This study uses time series and machine learning techniques to forecast Saudi Arabia’s refined petroleum output; a significant player in the global energy market. Using data from 1962 to 2022 acquired from the Ministry of Energy, Kingdom of Saudi Arabia, the study evaluates the forecasting performance of different models such as Facebook Prophet, Long Short-Term Memory (LSTM), Gaussian Process (GP), and Auto-Regressive Integrated Moving Average Model(ARIMA) based on metrics including Root mean squared error (RMSE), mean absolute percentage error (MAPE), relative absolute error (RAE), Akaike Information Criterion (AIC), and training time. The study results demonstrate that traditional time series models like ARIMA consistently exhibit superior prediction accuracy, whereas machine learning models like LSTM and GP offer more flexibility but need more data. Conversely, Prophet Model performs poorly as it often overlooks complex patterns within the data. The finding of this research work highlights the need for appropriate methodology use and careful model selection in predictive modeling initiatives to provide decision-makers with relevant information in the energy business. Future research may look into ways to use ensemble modeling techniques and other exogenous factors to increase the accuracy of forecasts for petroleum production.