Bitcoin Transaction Forecasting with DeepNetwork Representation Learning

ABSTARCT :

Bitcoin and its decentralized computing paradigm for digital currency trading are one of the most disruptive technology in the 21st century. This paper presents a novel approach to developing a Bitcoin transaction forecast model, DL Forecast, by leveraging deep neural networks for learning Bitcoin transaction network representations. DL Forecast makes three original contributions. To maintain an effective transaction forecasting performance, we leverage the multiplicative model update (MMU) ensemble to combine prediction models built on different transaction features extracted from each corresponding Bitcoin transaction graph. Evaluated on real-world Bitcoin transaction data, we show that our spatial-temporal forecasting model is efficient with fast runtime and effective with forecasting accuracy over 60% and improves the prediction performance by 50% when compared to forecasting model built on the static graph baseline.

EXISTING SYSTEM :

? We first provide an introduction to the real-world Bitcoin transaction dataset and demonstrate three key features : reach ability pattern, transaction amount pattern, and dynamics. ? The DLForecast development is inspired by two orthogonal research threads: (1) Statistic characterization of the Bitcoin transaction dataset. (2) Graph Mining. ? The development of our transaction forecasting DNN model consists of three main tasks.

DISADVANTAGE :

? Some of the transaction patterns in the present days are quite different from those of 5 years or 10 years ago. ? How to capture the up-to-date transaction patterns for accurate analysis and prediction on demand is a challenging problem. ? Bitcoin transaction addresses(accounts) have a short life span, and those transaction shappened in the past will have a very limited impact on future transactions, and such influence also decays overtime.

PROPOSED SYSTEM :

? In this paper, we present DLForecast, a Bitcoin transaction forecasting system, by leveraging deep network representation learning. ? Our goal is to predict transaction relationships among accounts on the Bitcoin network. ? This is the first paper applying DNN models on forecasting Bitcoin transactions using the real-world Bitcoin transaction data. ? The ensemble ensures the stable yet competitive performance of the proposed Bitcoin transaction forecasting system.

ADVANTAGE :

? The ensemble ensures the stable yet competitive performance of the proposed Bitcoin transaction forecasting system. ? We evaluate the forecasting performance of Bitcoin transactions using accuracy and f1-score. ? the transaction pattern changes over time and the forecasting performance relies heavily on the data itself, no single method can outperform all others.