دانلود مقاله پیش بینی قیمت بیت کوین با استفاده از یادگیری ماشین
چکیده
1. مقدمه
2. مطالعات مرتبط
3. روش شناسی
4. پیاده سازی
5.ارزیابی
6. نتیجه گیری و مطالعات آینده
منابع
Abstract
1.Introduction
2.Related Work
3.Methodology
4.Implementation
5.Evaluation
6.Conclusion and Future Work
References
چکیده
هدف این مقاله این است که مشخص شود با چه دقتی می توان جهت قیمت بیت کوین به دلار را پیش بینی کرد. داده‌های قیمت از شاخص قیمت بیت‌کوین گرفته شده است. این کار با درجات مختلف موفقیت از طریق اجرای یک شبکه عصبی بازگشتی بهینه بیزی (RNN) و یک شبکه حافظه کوتاه مدت (LSTM) به دست می‌آید. LSTM به بالاترین دقت طبقه بندی 52% و RMSE 8% دست می یابد. مدل محبوب ARIMA برای پیش بینی سری های زمانی به عنوان مقایسه با مدل های یادگیری عمیق پیاده سازی شده است. همانطور که انتظار می رود، روش های یادگیری عمیق غیرخطی بهتر از پیش بینی ARIMA هستند که عملکرد ضعیفی دارند. در نهایت، هر دو مدل یادگیری عمیق بر روی هر دو GPU و CPU با زمان آموزش در GPU بهتر از اجرای CPU تا 67.7٪ مقایسه شده اند.
توجه! این متن ترجمه ماشینی بوده و توسط مترجمین ای ترجمه، ترجمه نشده است.
Abstract
The goal of this paper is to ascertain with what accuracy the direction of Bitcoin price in USD can be predicted. The price data is sourced from the Bitcoin Price Index. The task is achieved with varying degrees of success through the implementation of a Bayesian optimised recurrent neural network (RNN) and a Long Short Term Memory (LSTM) network. The LSTM achieves the highest classification accuracy of 52% and a RMSE of 8%. The popular ARIMA model for time series forecasting is implemented as a comparison to the deep learning models. As expected, the non-linear deep learning methods outperform the ARIMA forecast which performs poorly. Finally, both deep learning models are benchmarked on both a GPU and a CPU with the training time on the GPU outperforming the CPU implementation by 67.7%.
Introduction
Bitcoin [1] is the worlds’ most valuable cryptocurrency and is traded on over 40 exchanges worldwide accepting over 30 different currencies. It has a current market capitalization of 9 billion USD according to https://www.blockchain.info/ and sees over 250,000 transactions taking place per day. As a currency, Bitcoin offers a novel opportunity for price prediction due its relatively young age and resulting volatility, which is far greater than that of fiat currencies [2]. It is also unique in relation to traditional fiat currencies in terms of its open nature; no complete data exists regarding cash transactions or money in circulation for fiat currencies.
Conclusion
Deep learning models such as the RNN and LSTM are evidently effective for Bitcoin prediction with the LSTM more capable for recognising longer-term dependencies. However, a high variance task of this nature makes it difficult to transpire this into impressive validation results. As a result it remains a difficult task. There is a fine line between overfitting a model and preventing it from learning sufficiently. Dropout is a valuable feature to assist in improving this. However, despite using Bayesian optimisation to optimize the selection of dropout it still couldn’t guarantee good validation results. Despite the metrics of sensitivity, specificity and precision indicating good performance, the actual performance of the ARIMA forecast based on error was significantly worse than the neural network models. The LSTM outperformed the RNN marginally, but not significantly. However, the LSTM takes considerably longer to train.