无法导入graph_utils等模块，求比特币异常检测相关安装方法

Hey there! I totally get the frustration when you're following a research paper and hit these import errors—let's break down how to fix this.

First off, these modules (graph_utils, custom_lstm_cell, evolve_graph_network) are not public Python packages you can install via pip. They're custom code that comes with the research paper you're referencing. Here's what you need to do:

• Locate the paper's official code repository
  Head back to the paper's page (like arXiv, IEEE Xplore, or the author's personal website). Most modern research papers include a link to their GitHub/GitLab repo in the footer or "Code Availability" section. Clone or download the entire code repository to your local machine, and unzip it if needed.

• Make sure the module files are in Python's search path
  Check the downloaded repository for the files graph_utils.py, custom_lstm_cell.py, and evolve_graph_network.py (or a subfolder containing them):

  1. If these files are in the same folder as your Jupyter Notebook/Python script, your import statements should work immediately.
  2. If they're in a subdirectory (e.g., src/ or models/), add that path to Python's sys path before importing:

     import sys
     sys.path.append("./path/to/module/folder")  # Replace with the actual path to the subdirectory
     import graph_utils
     import custom_lstm_cell
     import evolve_graph_network as egcn
     

• Install required dependencies
  These custom modules likely rely on specific versions of ML libraries (like TensorFlow/PyTorch, NetworkX, numpy). Check the repo's README.md or requirements.txt file, then install the dependencies with:

  pip install -r requirements.txt
  

  If there's no requirements file, refer to the paper's "Experimental Setup" section to install the necessary libraries matching the versions the authors used.

If you can't find the paper's code repo at all, you could try reaching out to the authors directly (most include contact info in the paper) or attempt to reimplement the modules based on the paper's descriptions—but this is more advanced, so prioritizing finding the official code is your best bet as a beginner.

内容的提问来源于stack exchange，提问作者Nikhil Kumar Macharla