How-To Guides

How to Upload a Dataset

Problem: You need to upload agricultural data for analysis and model training.

Solution:

1. Navigate to the Home page
2. Click on "Finding Similar Farmers" function
3. In the left panel, use the upload form
4. Enter a dataset name
5. Select a CSV file with your agricultural data
6. Click "Upload Dataset"

Troubleshooting:

• Ensure your CSV file is properly formatted
• Check that the file size is within limits
• Verify that the dataset name is unique

How to Train a Collaborative Model

Problem: You want to train a machine learning model using data from multiple farmers while preserving privacy.

Solution:

1. Navigate to "Collaborative Machine Learning"
2. Select a dataset from your uploaded datasets
3. Choose model parameters:
   • Model Name
   • Model Type (Dense, Conv1D, LSTM)
   • Model Visibility (Public/Private)
4. Review selected collaborators
5. Click "Start Training"

Advanced Tips:

• Use different model types for different data characteristics
• Consider privacy settings based on your data sensitivity
• Monitor training progress through the interface

How to Analyze Model Privacy Risks

Problem: You need to assess the privacy risks associated with your trained models.

Solution:

1. Go to the Model Repository
2. Find your model in the list
3. Click the red privacy risk icon (GppMaybeIcon)
4. Review the risk analysis sections:
   • Model Logs
   • Model Activity
   • Risk Analysis (Membership Inference Attack accuracy)

Troubleshooting:

• If risk analysis is not available, ensure the model training completed successfully
• Check that the model has sufficient data for analysis