【开源实习】在arxiv上发表基于MindSpore的原生论文（28）

research/arxiv_papers/traffic_prediction_mstim/traffic_prediction_mstim/README.md

@@ -0,0 +1,70 @@
+
+# Traffic Flow Prediction using MSTIM Model (MindSpore)
+
+This repository implements a **multi-scale temporal information modeling (MSTIM)** based deep learning model for traffic flow prediction, developed using the **MindSpore** framework. The model combines LSTM, CNN, and attention-style temporal aggregation to capture both short-term and long-term patterns in traffic data.
+
+## 📁 Project Structure
+
+```
+traffic_prediction_mstim/
+├── data/                   # Contains the traffic dataset (e.g., Traffic.csv)
+├── models/
+│   ├── mstim.py            # MSTIM module definition
+│   └── model.py            # TrafficFlowPredictor model
+├── utils/
+│   ├── preprocess.py       # Data preprocessing functions
+│   ├── dataset.py          # Dataset wrapper for MindSpore
+│   └── metrics.py          # Evaluation metric computation
+├── train.py                # Main training and evaluation script
+├── config.py               # (Optional) Configuration settings
+├── requirements.txt        # Project dependencies
+└── README.md               # Project description
+```
+
+## 📊 Dataset
+
+We use the [Metro Interstate Traffic Volume Dataset](https://archive.ics.uci.edu/ml/datasets/Metro+Interstate+Traffic+Volume), which contains hourly traffic volume data from a highway in Minnesota, USA. The dataset includes weather, date-time, and traffic volume features.
+
+- Format: CSV
+- Time range: 2012 to 2018
+- Target: `traffic_volume` (normalized)
+
+## 🚀 How to Run
+
+1. **Install dependencies**  
+   (We recommend using a virtual environment)
+
+   ```bash
+   pip install -r requirements.txt
+   ```
+
+2. **Place dataset**  
+   Download and place the CSV data file in the `data/` directory. You can rename it as `Traffic.csv` or modify the path in `train.py`.
+
+3. **Train and Evaluate Model**
+
+   ```bash
+   python train.py
+   ```
+
+4. **Model Output**  
+   - Checkpoint saved as `traffic_flow_predictor.ckpt`
+   - Evaluation metrics: MAE, MSE, RMSE
+
+## 🧠 Model Highlights
+
+- **MSTIM Module**: Combines multiple 1D convolution layers with Bi-LSTM for multi-scale sequence modeling.
+- **LSTM Backbone**: Learns temporal dependencies in the input sequence.
+- **Custom Preprocessing**: Includes feature standardization and weather condition encoding.
+
+## 📦 Dependencies
+
+See [`requirements.txt`](./requirements.txt) for a complete list of required packages.
+
+## 📄 License
+
+MIT License
+
+---
+
+Created by your AI team using the MindSpore deep learning framework.

research/arxiv_papers/traffic_prediction_mstim/traffic_prediction_mstim/config.py

@@ -0,0 +1,19 @@
+
+# Configuration settings for Traffic Flow Prediction Model
+
+# Data settings
+DATA_PATH = "data/Traffic.csv"
+WINDOW_SIZE = 24
+
+# Model settings
+LSTM_HIDDEN_SIZE = 64
+LSTM_LAYERS = 1
+MSTIM_OUT_CHANNELS = 32
+
+# Training settings
+BATCH_SIZE = 32
+NUM_EPOCHS = 10
+LEARNING_RATE = 0.001
+
+# MindSpore context settings
+DEVICE_TARGET = "CPU"  # Change to "GPU" if GPU is available

research/arxiv_papers/traffic_prediction_mstim/traffic_prediction_mstim/requirements.txt

@@ -0,0 +1,6 @@
+
+mindspore>=1.10.0
+numpy
+pandas
+scikit-learn
+tqdm