Feat/add docker support

Dockerfile

@@ -0,0 +1,26 @@
+# 1. Base Image:
+#    - PyTorch 1.11.0
+#    - CUDA 11.3 (The corresponding version for torch 1.11.0)
+#    - Python 3.9.12 (Closest available to 3.9.13)
+FROM pytorch/pytorch:1.11.0-cuda11.3-cudnn8-runtime
+
+# 2. Set Working Directory
+WORKDIR /app
+
+# 3. (CRITICAL!) Package Installation
+#    We are avoiding installing directly from a local 'requirements.txt'
+#    because it often causes Python version conflicts (specifically with Python 3.9)
+#    due to pre-compiled binaries or version lock issues.
+#    Instead, we directly install the core packages needed for the project,
+#    allowing pip to automatically select compatible versions for this environment.
+RUN pip install --no-cache-dir \
+    numpy \
+    pandas \
+    scikit-learn \
+    matplotlib
+
+# 4. Copy all project files
+COPY . .
+
+# 5. Default command: Start bash shell
+CMD ["/bin/bash"]

README.md

@@ -22,6 +22,19 @@ pip install torch==1.11.0
 pip install -r requirements.txt
 ```
 
+- If you prefer Docker (requires Docker Engine and NVIDIA Container Toolkit):
+```bash
+# 1. Build the image (run from the directory containing the Dockerfile)
+docker build -t d3r .
+
+# 2. Run the experiment
+#    (Ensure your downloaded datasets are in this directory for the mount to work)
+docker run --rm -it --gpus all \
+    -v .:/app/data \
+    d3r \
+    /bin/bash
+```
+
 2. Download data. You can obtain two benchmarks from [Google Cloud](https://drive.google.com/drive/folders/1UJ6SGfb6h-9R0L18FLDXpISKh1nhaqWA?usp=sharing). The datasets are well pre-processed. For the SWaT dataset, you can apply for it by following its [official tutorial](https://itrust.sutd.edu.sg/itrust-labs_datasets/dataset_info/). We unify the SWaT dataset to minute granularity and retain only continuous metrics:
 
    ```