A survey of deep learning applications in cryptocurrency

Area of Science:

• Multidisciplinary research at the intersection of artificial intelligence and financial technology.
• Focus on the burgeoning field of deep learning applications within the cryptocurrency domain.

Background:

• Cryptocurrencies have evolved significantly, with major digital assets shaping the financial landscape.
• Deep learning models are increasingly utilized across various financial applications, demonstrating their potential.

Purpose of the Study:

• To provide a comprehensive review of deep learning methods applied to cryptocurrency research.
• To synthesize existing literature on deep learning in cryptocurrency for diverse modeling tasks.
• To identify key challenges and future research avenues in this interdisciplinary area.

Main Methods:

• Systematic literature review of deep learning models (CNNs, RNNs, DBNs, Deep RL) and their use in cryptocurrency.
• Overview of cryptocurrency history and representative digital currencies.
• Analysis of studies across various cryptocurrency research tasks: price prediction, portfolio construction, bubble analysis, abnormal trading, trading regulations, and initial coin offerings.

Main Results:

• Deep learning models are applied to a wide range of cryptocurrency research problems, from market prediction to regulatory analysis.
• Evaluation of studies based on modeling approaches, data, results, and innovations highlights current trends and limitations.
• Identified gaps in research concerning specific deep learning architectures and their efficacy in novel cryptocurrency applications.

Conclusions:

• Deep learning offers significant potential for advancing cryptocurrency research across multiple domains.
• Further research is needed to explore advanced deep learning techniques and their practical implications in the cryptocurrency market.
• Future work should focus on developing robust models for complex tasks like regulatory compliance and detecting market manipulation.