Démarrage rapide XGBoost

In this federated learning tutorial, we will learn how to train a simple XGBoost classifier on Higgs dataset using Flower and XGBoost. It is recommended to create a virtual environment and run everything within a virtualenv.

Let’s use flwr new to create a complete Flower+XGBoost project. It will generate all the files needed to run, by default with the Simulation Engine, a federation of 10 nodes using FedXgbBagging strategy. The dataset will be partitioned using Flower Dataset’s IidPartitioner.

FedXgbBagging (bootstrap aggregation) is an ensemble method that improves stability and accuracy in machine learning, here applied to XGBoost in FL. Each client generates a bootstrap sample by subsampling its data and trains a tree per round, which is then aggregated by the server and added to the global model.

Now that we have a rough idea of what this example is about, let’s get started. First, install Flower in your new environment:

# In a new Python environment
$ pip install flwr

Then, run the command below. You will be prompted to select one of the available templates (choose XGBoost), give a name to your project, and enter your developer name:

$ flwr new

After running it you’ll notice a new directory with your project name has been created. It should have the following structure:

<your-project-name>
├── <your-project-name>
│   ├── __init__.py
│   ├── client_app.py   # Defines your ClientApp
│   ├── server_app.py   # Defines your ServerApp
│   └── task.py         # Defines your data loading and utility functions
├── pyproject.toml      # Project metadata like dependencies and configs
└── README.md

If you haven’t yet installed the project and its dependencies, you can do so by:

# From the directory where your pyproject.toml is
$ pip install -e .

To run the project do:

# Run with default arguments
$ flwr run .

With default arguments, you will see output like this:

Loading project configuration...
Success
INFO :      Starting FedXgbBagging strategy:
INFO :              ├── Number of rounds: 3
INFO :              ├── ArrayRecord (0.00 MB)
INFO :              ├── ConfigRecord (train): (empty!)
INFO :              ├── ConfigRecord (evaluate): (empty!)
INFO :              ├──> Sampling:
INFO :              │       ├──Fraction: train (0.10) | evaluate ( 0.10)
INFO :              │       ├──Minimum nodes: train (2) | evaluate (2)
INFO :              │       └──Minimum available nodes: 2
INFO :              └──> Keys in records:
INFO :                      ├── Weighted by: 'num-examples'
INFO :                      ├── ArrayRecord key: 'arrays'
INFO :                      └── ConfigRecord key: 'config'
INFO :
INFO :
INFO :      [ROUND 1/3]
INFO :      configure_train: Sampled 2 nodes (out of 10)
INFO :      aggregate_train: Received 2 results and 0 failures
INFO :              └──> Aggregated MetricRecord: {}
INFO :      configure_evaluate: Sampled 2 nodes (out of 10)
INFO :      aggregate_evaluate: Received 2 results and 0 failures
INFO :              └──> Aggregated MetricRecord: {'auc': 0.7677505289821278}
INFO :
INFO :      [ROUND 2/3]
INFO :      configure_train: Sampled 2 nodes (out of 10)
INFO :      aggregate_train: Received 2 results and 0 failures
INFO :              └──> Aggregated MetricRecord: {}
INFO :      configure_evaluate: Sampled 2 nodes (out of 10)
INFO :      aggregate_evaluate: Received 2 results and 0 failures
INFO :              └──> Aggregated MetricRecord: {'auc': 0.7758267351298489}
INFO :
INFO :      [ROUND 3/3]
INFO :      configure_train: Sampled 2 nodes (out of 10)
INFO :      aggregate_train: Received 2 results and 0 failures
INFO :              └──> Aggregated MetricRecord: {}
INFO :      configure_evaluate: Sampled 2 nodes (out of 10)
INFO :      aggregate_evaluate: Received 2 results and 0 failures
INFO :              └──> Aggregated MetricRecord: {'auc': 0.7811659285552999}
INFO :
INFO :      Strategy execution finished in 132.88s
INFO :
INFO :      Final results:
INFO :
INFO :              Global Arrays:
INFO :                      ArrayRecord (0.195 MB)
INFO :
INFO :              Aggregated ClientApp-side Train Metrics:
INFO :              {1: {}, 2: {}, 3: {}}
INFO :
INFO :              Aggregated ClientApp-side Evaluate Metrics:
INFO :              { 1: {'auc': '7.6775e-01'},
INFO :                2: {'auc': '7.7583e-01'},
INFO :                3: {'auc': '7.8117e-01'}}
INFO :
INFO :              ServerApp-side Evaluate Metrics:
INFO :              {}
INFO :

Saving final model to disk...

You can also override the parameters defined in the [tool.flwr.app.config] section in the pyproject.toml like this:

# Override some arguments
$ flwr run . --run-config "num-server-rounds=5 params.eta=0.2"

What follows is an explanation of each component in the project you just created: configurations, dataset partitioning, defining the ClientApp, and defining the ServerApp.

The Configurations

We define all required configurations / hyper-parameters inside the pyproject.toml file:

[tool.flwr.app.config]
num-server-rounds = 3
fraction-train = 0.1
fraction-evaluate = 0.1
local-epochs = 1

# XGBoost parameters
params.objective = "binary:logistic"
params.eta = 0.1 # Learning rate
params.max-depth = 8
params.eval-metric = "auc"
params.nthread = 16
params.num-parallel-tree = 1
params.subsample = 1
params.tree-method = "hist"

The local-epochs represents the number of iterations for local tree boost. We use CPU for the training in default. One can assign it to a GPU by setting tree-method to gpu_hist. We use AUC as evaluation metric.

The Data

We will use Flower Datasets to easily download and partition the Higgs dataset. In this example, you’ll make use of the IidPartitioner to generate num_partitions partitions. You can choose from other partitioners available in Flower Datasets:

partitioner = IidPartitioner(num_partitions=num_clients)
fds = FederatedDataset(
    dataset="jxie/higgs",
    partitioners={"train": partitioner},
)
partition = fds.load_partition(partition_id, split="train")
partition.set_format("numpy")

# Train/test splitting
train_data, valid_data, num_train, num_val = train_test_split(
    partition, test_fraction=0.2, seed=42
)

# Reformat data to DMatrix for xgboost
train_dmatrix = transform_dataset_to_dmatrix(train_data)
valid_dmatrix = transform_dataset_to_dmatrix(valid_data)

We train/test split using the given partition (client’s local data), and reformat data to DMatrix for the xgboost package. The functions of train_test_split and transform_dataset_to_dmatrix are defined as below:

def train_test_split(partition, test_fraction, seed):
    """Split the data into train and validation set given split rate."""
    train_test = partition.train_test_split(test_size=test_fraction, seed=seed)
    partition_train = train_test["train"]
    partition_test = train_test["test"]

    num_train = len(partition_train)
    num_test = len(partition_test)

    return partition_train, partition_test, num_train, num_test


def transform_dataset_to_dmatrix(data):
    """Transform dataset to DMatrix format for xgboost."""
    x = data["inputs"]
    y = data["label"]
    new_data = xgb.DMatrix(x, label=y)
    return new_data

The ClientApp

The main changes we have to make to use XGBoost with Flower have to do with converting the ArrayRecord received in the Message into a XGBoost loadable binary object, and vice versa when generating the reply Message from the ClientApp. We can make use of the following conversions:

@app.train()
def train(msg: Message, context: Context):

    # Instantiate a XGBoost model
    bst = xgb.Booster(params=params)
    global_model = bytearray(msg.content["arrays"]["0"].numpy().tobytes())

    # Load global model into booster
    bst.load_model(global_model)

    # ...

    # Convert XGB object back into an ArrayRecord
    # Note: we store the model as the first item in a list into ArrayRecord,
    # which can be accessed using index ["0"].
    local_model = bst.save_raw("json")
    model_np = np.frombuffer(local_model, dtype=np.uint8)
    model_record = ArrayRecord([model_np])

The rest of the functionality is directly inspired by the centralized case. The ClientApp comes with three core methods (train, evaluate, and query) that we can implement for different purposes. For example: train to train the received model using the local data; evaluate to assess its performance of the received model on a validation set; and query to retrieve information about the node executing the ClientApp. In this tutorial we will only make use of train and evaluate.

Let’s see how the train method can be implemented. It receives as input arguments a Message from the ServerApp. By default it carries:

• an ArrayRecord with the arrays of the model to federate. By default they can be retrieved with key "arrays" when accessing the message content.

• a ConfigRecord with the configuration sent from the ServerApp. By default it can be retrieved with key "config" when accessing the message content.

The train method also receives the Context, giving access to configs for your run and node. The run config hyperparameters are defined in the pyproject.toml of your Flower App. The node config can only be set when running Flower with the Deployment Runtime and is not directly configurable during simulations.

# Flower ClientApp
app = ClientApp()


@app.train()
def train(msg: Message, context: Context) -> Message:
    # Load model and data
    partition_id = context.node_config["partition-id"]
    num_partitions = context.node_config["num-partitions"]
    train_dmatrix, _, num_train, _ = load_data(partition_id, num_partitions)

    # Read from run config
    num_local_round = context.run_config["local-epochs"]
    # Flatted config dict and replace "-" with "_"
    cfg = replace_keys(unflatten_dict(context.run_config))
    params = cfg["params"]

    global_round = msg.content["config"]["server-round"]
    if global_round == 1:
        # First round local training
        bst = xgb.train(
            params,
            train_dmatrix,
            num_boost_round=num_local_round,
        )
    else:
        bst = xgb.Booster(params=params)
        global_model = bytearray(msg.content["arrays"]["0"].numpy().tobytes())

        # Load global model into booster
        bst.load_model(global_model)

        # Local training
        bst = _local_boost(bst, num_local_round, train_dmatrix)

    # Save model
    local_model = bst.save_raw("json")
    model_np = np.frombuffer(local_model, dtype=np.uint8)

    # Construct reply message
    # Note: we store the model as the first item in a list into ArrayRecord,
    # which can be accessed using index ["0"].
    model_record = ArrayRecord([model_np])
    metrics = {
        "num-examples": num_train,
    }
    metric_record = MetricRecord(metrics)
    content = RecordDict({"arrays": model_record, "metrics": metric_record})
    return Message(content=content, reply_to=msg)

At the first round, we call xgb.train() to build up the first set of trees. From the second round, we load the global model sent from server to new build Booster object, and then update model weights on local training data with function _local_boost as follows:

def _local_boost(self, bst_input):
    # Update trees based on local training data.
    for i in range(self.num_local_round):
        bst_input.update(self.train_dmatrix, bst_input.num_boosted_rounds())

    # Bagging: extract the last N=num_local_round trees for sever aggregation
    bst = bst_input[
        bst_input.num_boosted_rounds()
        - self.num_local_round : bst_input.num_boosted_rounds()
    ]

    return bst

Given num_local_round, we update trees by calling bst_input.update method. After training, the last N=num_local_round trees will be extracted to send to the server.

The @app.evaluate() method would be near identical with two exceptions: (1) the model is not locally trained, instead it is used to evaluate its performance on the locally held-out validation set; (2) including the model in the reply Message is no longer needed because it is not locally modified.

The ServerApp

To construct a ServerApp, we define its @app.main() method. This method receives as input arguments:

• a Grid object that will be used to interface with the nodes running the ClientApp to involve them in a round of train/evaluate/query or other.

• a Context object that provides access to the run configuration.

In this example we use the FedXgbBagging strategy. Then, we initialize an empty global model as the XGBoost model will be initialized on client side in the first round. After that, the execution of the strategy is launched when invoking its start method. To it we pass:

• the Grid object.

• an ArrayRecord carrying a randomly initialized model that will serve as the global

  model to federate.

• the num_rounds parameter specifying how many rounds to perform.

# Create ServerApp
app = ServerApp()


@app.main()
def main(grid: Grid, context: Context) -> None:
    # Read run config
    num_rounds = context.run_config["num-server-rounds"]
    fraction_train = context.run_config["fraction-train"]
    fraction_evaluate = context.run_config["fraction-evaluate"]
    # Flatted config dict and replace "-" with "_"
    cfg = replace_keys(unflatten_dict(context.run_config))
    params = cfg["params"]

    # Init global model
    # Init with an empty object; the XGBooster will be created
    # and trained on the client side.
    global_model = b""
    # Note: we store the model as the first item in a list into ArrayRecord,
    # which can be accessed using index ["0"].
    arrays = ArrayRecord([np.frombuffer(global_model, dtype=np.uint8)])

    # Initialize FedXgbBagging strategy
    strategy = FedXgbBagging(
        fraction_train=fraction_train,
        fraction_evaluate=fraction_evaluate,
    )

    # Start strategy, run FedXgbBagging for `num_rounds`
    result = strategy.start(
        grid=grid,
        initial_arrays=arrays,
        num_rounds=num_rounds,
    )

    # Save final model to disk
    bst = xgb.Booster(params=params)
    global_model = bytearray(result.arrays["0"].numpy().tobytes())

    # Load global model into booster
    bst.load_model(global_model)

    # Save model
    print("\nSaving final model to disk...")
    bst.save_model("final_model.json")

Note the start method of the strategy returns a Result object. This object contains all the relevant information about the FL process, including the final model weights as an ArrayRecord, and federated training and evaluation metrics as MetricRecords.

Congratulations! You’ve successfully built and run your first federated learning system.

Note

Check the source code of the extended version of this tutorial in examples/xgboost-quickstart in the Flower GitHub repository.