Scholarlink

The analysis of cryptocurrency markets is a very tough task due to their extreme volatility. Our work presents a hybrid model, using LSTM neural networks combined with the Dragonfly Algorithm, to perform hyper-parameter optimization when predicting cryptocurrency prices. Due to their nature, LSTM models are normally good at modeling time-series dependencies, while DA is intended for use to optimize the hyper-parameters of the model to improve prediction accuracy. We will apply the methodology to a dataset of several cryptocurrencies, preprocessed by features like open, close, high, low, and volume. The hybrid model demonstrates improved performance in predicting price trends compared to traditional methods, making it a potentially effective solution for addressing the challenges faced in predicting cryptocurrency prices.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Cryptocurrency Price Prediction Using LSTM and DragonFly Optimization Algorithm

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Padmavathi,
  • C. J. Niraimathi,
  • Smruti Padhi

摘要

The analysis of cryptocurrency markets is a very tough task due to their extreme volatility. Our work presents a hybrid model, using LSTM neural networks combined with the Dragonfly Algorithm, to perform hyper-parameter optimization when predicting cryptocurrency prices. Due to their nature, LSTM models are normally good at modeling time-series dependencies, while DA is intended for use to optimize the hyper-parameters of the model to improve prediction accuracy. We will apply the methodology to a dataset of several cryptocurrencies, preprocessed by features like open, close, high, low, and volume. The hybrid model demonstrates improved performance in predicting price trends compared to traditional methods, making it a potentially effective solution for addressing the challenges faced in predicting cryptocurrency prices.