26 February 2025 in IEEE Access
Algorithmic stock trading has improved tremendously, with Reinforcement Learning (RL) algorithms being more adaptable than classic approaches like mean reversion and momentum. However, challenges remain in adequately depicting market events and generating suitable rewards to influence the trading decisions of an agent in a dynamic environment. This study proposed an improved stock market trading framework termed the RMS-Driven Deep Reinforcement Learning (DRL) model for optimal portfolio management. The research attempts to give a more comprehensive view of the market and, as a result, enhance trading decisions by including consumer information and incorporating news sentiments into the model, in addition to data from typical earnings reports. More specifically, three kinds of DRL models are presented, combined with data from stock earnings reports, Max Drawdown rewards, and sentiment indicators (RMS), known as PPO_RMS, A2C_RMS, and DDPG_RMS, respectively. The findings of this research indicate that the integrated model, mainly the DDPG_RMS effectively outperforms the baseline ^DJI index in many risk-return analyses in ratios showing better risk management, and profitability. The proposed stock trading model generates a maximum cumulative return of 27% with a Sharpe ratio of 0.66, showing an appropriate trade-off between risk and return. This approach, which incorporates sentiment analysis and Max Drawdown rewards, significantly enhances the model’s performance in adapting to changing market conditions. Therefore, the results emphasize the appropriateness of integrating sentiment indices with traditional financial data to enhance a trader’s performance while also offering essential information to aid in the development of improved trading tactics in continuously changing financial markets.
1 May 2025 in Expert Systems with Applications
1 January 2026 in IEEE Access
4 December 2025 in ICT Express
12 March 2026 in IEEE Access
Deepfakes are an increasing danger to the reliability of digital media, with several detection algorithms developed to identify manipulated content. Although detection approaches continue to improve, proactive defenses that protect images by embedding imperceptible adversarial perturbations before distribution, thereby preventing deepfake generation at the source, offer a more robust complementary solution. However, existing proactive adversarial defenses against deepfake generators suffer from three key limitations: poor imperceptibility, limited texture-level targeting, and a lack of control over low-level structure perturbation. To address this, a novel patch-wise texture-aware adversarial attack is proposed as a proactive defense that disrupts deepfake generation. This is because adversarial images are generated through the addition of perturbations to the original facial images, so maintaining visual imperceptibility is a critical requirement; thus, our defense explicitly focuses on limiting perturbations to low-energy texture aspects that do not significantly affect human perception. The proposed method leverages tail-mode singular vectors extracted via patch-wise Singular Value Decomposition (SVD) to amplify low-energy texture components, i.e., producing adversarial perturbations that remain visually imperceptible while degrading the quality of generated deepfake images. Additionally, we introduce a variant that performs gradient projection into the SVD-tail subspace, ensuring that updates stay confined to subtle texture patterns and avoid high-frequency noise. The findings show that the proposed method is a stealthier attack that disrupts deepfake generations. Experimental results show that the proposed method achieves high attack success rates while preserving perceptual similarity, outperforming baselines in terms of imperceptibility and effectiveness. This work demonstrates a novel pathway toward proactive and imperceptible defenses against deepfake generation.