Papers by Wary Hossain Rabby
International Journal of Research and Scientific Innovation, 2025
We're researching how our thesis can help guess stock prices and suggest smart investment moves. ... more We're researching how our thesis can help guess stock prices and suggest smart investment moves. We start by checking if the current stock prices are right, looking at both the percentage and money differences. We also predict the prices tomorrow, showing the real-time and guessed numbers and explaining how much they differ. After that, we give practical advice in three categories: Sell, Hold, and Buy, so people can make smart choices. We also look at what happens if the stock prices are guessed wrong and how it affects people's investment portfolios. a Contribution to Knowledge and Statement of Significance 1. This thesis significantly advances financial knowledge by: 2. Enhances understanding of stock price accuracy through machine learning analysis. 3. Provides valuable insights into next-day price predictions with real-time comparisons. 4. Contributes practical recommendations for buying shares at critical stages. 5. Highlights the significant impact of incorrect share prices on investment portfolios. Thesis Organization We have divided our thesis into five chapters. In Chapter 1, Introduction and some related works are reviewed. Chapter 2 gives a Literature review that indicates the related work to our thesis. Chapter 3, explains our research methodology and the mechanism of how we work. Chapter 4, presents an analysis of the results of our thesis applied to our dataset and discussion. Chapter 5, represents a Conclusion of our thesis. LITERATURE REVIEW Filtering an audio signal with an all-pass filter does not usually have a major effect on the signal's timbre. The all-pass filter does not change the frequency content of the signal, but only introduces a phase shift or delay. Audibility of the phase distortion caused by an all-pass filter in a sound reproduction system has been a topic of many studies, see, e.g., [1], [2]. In this paper, we investigate audio effects processing using high-order all-pass filters that consist of many cascaded low-order all-pass filters. These filters have long chirp-like impulse responses. When audio and music signals are processed with such a filter, remarkable changes are obtained that are similar to the spectral delay effect [3], [4].
International Journal of Science and Research archive, 2025
One well-liked method for condensing massive language models (LLMs) into smaller, faster, more ef... more One well-liked method for condensing massive language models (LLMs) into smaller, faster, more effective versions without sacrificing performance is knowledge distillation (KD). However, it is no longer feasible to run distillation on a single device as LLMs grow to hundreds of billions of parameters; it is simply too computationally demanding. In this paper, we investigate how to leverage multi-GPU configurations to make KD scale. To overcome communication bottlenecks and accelerate training, our method combines adaptive gradient compression with tensor, pipeline, and data parallelism. Tested on transformer-based LLMs, our framework maintains strong accuracy for both understanding and generation tasks, reduces communication overhead by 27%, and provides up to 3.4× faster training than single-GPU baselines.
Uploads
Papers by Wary Hossain Rabby