feltlabs 0.5.2

FELT python package intended for running federated learning on Ocean protocol.

This code is intended to work closely with Ocean protocol. Algorithms from this code should run on ocean provider. Training local models and aggregating them into global model.

Entry commands:

felt-train
felt-aggregate
felt-export

Common Usage

After installing this library you can load models trained using FELT as:

from feltlabs.model import load_model

# Load scikit-learn model
model = load_model("final-model.json")

# Data shape must be: (number_of_samples, number_of_features)
data = [
  [1980, 2, 2, 2, 0, 0],
  [1700, 3, 2, 3, 1, 1],
  [2100, 3, 2, 3, 1, 0],
]

result = model.predict(data)
print(result)
# Use following line for analytics as mean, std...
# result = model.predict(None)

Command: felt-export

You can use felt-export for converting trained models into pickle object: Resulting file will then contain a pickled object of scikit-learn model.

felt-export --input "final-model-House Prices.json" --output "model.pkl"

Then you can use the created file as follows:

import pickle

with open('model.pkl', 'rb') as f:
    model = pickle.load(object, f)
    
# See the above code example for data definition
model.predict(data)