Get started with Singularity Machine Learning

Note

Qiskit Functions are an experimental feature available only to IBM Quantum® Premium Plan, Flex Plan, and On-Prem (via IBM Quantum Platform API) Plan users. They are in preview release status and subject to change.

For an overview of this function, see the Singularity Machine Learning documentation home.

The code on this page was developed using the following requirements. We recommend using these versions or newer.
```
scikit-learn~=1.8.0
```

Set up

Authenticate using your IBM Quantum Platform API key, and select the Qiskit Function as follows:

from qiskit_ibm_catalog import QiskitFunctionsCatalog

catalog = QiskitFunctionsCatalog(channel="ibm_quantum_platform")

# Verify that you have access to the function
catalog.list()

Output:

[QiskitFunction(qunova/hivqe-chemistry),
 QiskitFunction(global-data-quantum/quantum-portfolio-optimizer),
 QiskitFunction(algorithmiq/tem),
 QiskitFunction(qedma/qesem),
 QiskitFunction(multiverse/singularity),
 QiskitFunction(ibm/circuit-function),
 QiskitFunction(q-ctrl/optimization-solver),
 QiskitFunction(colibritd/quick-pde),
 QiskitFunction(q-ctrl/performance-management),
 QiskitFunction(kipu-quantum/iskay-quantum-optimizer)]

# load function
singularity = catalog.load("multiverse/singularity")

Examples

Classify a dataset

In this example, you'll use the "Singularity Machine Learning - Classification" function to classify a dataset consisting of two interleaving, moon-shaped half-circles. The dataset is synthetic, two-dimensional, and labeled with binary labels. It is created to be challenging for algorithms such as centroid-based clustering and linear classification.

Through this process, you'll learn how to create the classifier, fit it to the training data, use it to predict on the test data, and delete the classifier when you're finished.

Before starting, you need to install scikit-learn. Install it using the following command:

python3 -m pip install scikit-learn

Perform the following steps:

Create the synthetic dataset using the make_moons function from scikit-learn.
Upload the generated synthetic dataset to the shared data directory.
Create the quantum-enhanced classifier using the create action.
Enlist your classifiers using the list action.
Train the classifier on the train data using the fit action.
Use the trained classifier to predict on the test data using the predict action.
Delete the classifier using the delete action.
Clean up after you're done.

Step 1. Import the necessary modules and generate the synthetic dataset, then split it into training and test datasets.

# import the necessary modules for this example
import os
import tarfile
import numpy as np

# Import the make_moons and the train_test_split functions from scikit-learn
# to create a synthetic dataset and split it into training and test datasets
from sklearn.datasets import make_moons
from sklearn.model_selection import train_test_split

# generate the synthetic dataset
X, y = make_moons(n_samples=10000)

# split the data into training and test datasets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)

# print the first 10 samples of the training dataset
print("Features:", X_train[:10, :])
print("Targets:", y_train[:10])

Output:

Features: [[ 0.84757037 -0.48831433]
 [ 0.98132552  0.19235443]
 [-0.71626723  0.6978261 ]
 [ 1.18957848 -0.48186557]
 [ 0.52118982 -0.37791846]
 [ 0.81115408  0.58483251]
 [ 0.48706462  0.87336593]
 [-0.81880144  0.57407682]
 [ 1.67335408 -0.23932015]
 [ 0.50181306  0.8649761 ]]
Targets: [1 0 0 1 1 0 0 0 1 0]

Step 2. Save the labeled training and test datasets on your local disk, and then upload them to the shared data directory.

def make_tarfile(file_path, tar_file_name):
    with tarfile.open(tar_file_name, "w") as tar:
        tar.add(file_path, arcname=os.path.basename(file_path))


# save the training and test datasets on your local disk
np.save("X_train.npy", X_train)
np.save("y_train.npy", y_train)
np.save("X_test.npy", X_test)
np.save("y_test.npy", y_test)

# create tar files for the datasets
make_tarfile("X_train.npy", "X_train.npy.tar")
make_tarfile("y_train.npy", "y_train.npy.tar")
make_tarfile("X_test.npy", "X_test.npy.tar")
make_tarfile("y_test.npy", "y_test.npy.tar")

# upload the datasets to the shared data directory
catalog.file_upload("X_train.npy.tar", singularity)
catalog.file_upload("y_train.npy.tar", singularity)
catalog.file_upload("X_test.npy.tar", singularity)
catalog.file_upload("y_test.npy.tar", singularity)

# view/enlist the uploaded files in the shared data directory
print(catalog.files(singularity))

Output:

['X_test.npy.tar', 'X_train.npy.tar', 'y_test.npy.tar', 'y_train.npy.tar']

Step 3. Create a quantum-enhanced classifier using the create action.

job = singularity.run(
    action="create",
    name="my_classifier",
    num_learners=10,
    learners_types=[
        "DecisionTreeClassifier",
        "KNeighborsClassifier",
    ],
    learners_proportions=[0.5, 0.5],
    learners_options=[{}, {}],
    regularization=0.01,
    weight_update_method="logarithmic",
    sample_scaling=True,
    optimizer_options={"simulator": True},
    voting="soft",
    prob_threshold=0.5,
)

print(job.result())

Output:

{'status': 'ok', 'message': 'Classifier created.', 'data': {}, 'metadata': {'resource_usage': {}}}

# list available classifiers using the list action
job = singularity.run(action="list")

print(job.result())

# you can also find your classifiers in the shared data directory with a *.pkl.tar extension
print(catalog.files(singularity))

Output:

{'status': 'ok', 'message': 'Classifiers listed.', 'data': {'classifiers': ['my_classifier']}, 'metadata': {'resource_usage': {}}}

['X_test.npy.tar', 'X_train.npy.tar', 'my_classifier.pkl.tar', 'y_test.npy.tar', 'y_train.npy.tar']

Step 4. Train the quantum-enhanced classifier using the fit action.

job = singularity.run(
    action="fit",
    name="my_classifier",
    X="X_train.npy",  # you do not need to specify the tar extension
    y="y_train.npy",  # you do not need to specify the tar extension
)

print(job.result())

Output:

{'status': 'ok', 'message': 'Classifier fitted.', 'data': {}, 'metadata': {'resource_usage': {'RUNNING: MAPPING': {'CPU_TIME': 13.655871629714966}, 'RUNNING: WAITING_QPU': {'CPU_TIME': 54.688621282577515}, 'RUNNING: POST_PROCESSING': {'CPU_TIME': 56.92286920547485}, 'RUNNING: EXECUTING_QPU': {'QPU_TIME': 57.92738223075867}}}}

Step 5. Obtain predictions and probabilities from the quantum-enhanced classifier using the predict action.

job = singularity.run(
    action="predict",
    name="my_classifier",
    X="X_test.npy",  # you do not need to specify the tar extension
)

result = job.result()

print("Action result status: ", result["status"])
print("Action result message: ", result["message"])
print("Predictions (first five results):", result["data"]["predictions"][:5])
print(
    "Probabilities (first five results):", result["data"]["probabilities"][:5]
)

Output:

Action result status:  ok
Action result message:  Classifier predicted.
Predictions (first five results): [0, 0, 1, 0, 1]
Probabilities (first five results): [[1.0, 0.0], [1.0, 0.0], [0.0, 1.0], [1.0, 0.0], [0.0, 1.0]]

Step 6. Delete the quantum-enhanced classifier using the delete action.

job = singularity.run(
    action="delete",
    name="my_classifier",
)

# or you can delete from the shared data directory
# catalog.file_delete("my_classifier.pkl.tar", singularity)

print(job.result())

Output:

{'status': 'ok', 'message': 'Classifier deleted.', 'data': {}, 'metadata': {'resource_usage': {}}}

Step 7. Clean up local and shared data directories.

# delete the numpy files from your local disk
os.remove("X_train.npy")
os.remove("y_train.npy")
os.remove("X_test.npy")
os.remove("y_test.npy")

# delete the tar files from your local disk
os.remove("X_train.npy.tar")
os.remove("y_train.npy.tar")
os.remove("X_test.npy.tar")
os.remove("y_test.npy.tar")

# delete the tar files from the shared data
catalog.file_delete("X_train.npy.tar", singularity)
catalog.file_delete("y_train.npy.tar", singularity)
catalog.file_delete("X_test.npy.tar", singularity)
catalog.file_delete("y_test.npy.tar", singularity)

Output:

'Requested file was deleted.'

`create_fit_predict` example

The following example demonstrates the create_fit_predict action.

# Import QiskitFunctionsCatalog to load the
# "Singularity Machine Learning - Classification" function by Multiverse Computing
from qiskit_ibm_catalog import QiskitFunctionsCatalog

# Import the make_moons and the train_test_split functions from scikit-learn
# to create a synthetic dataset and split it into training and test datasets
from sklearn.datasets import make_moons
from sklearn.model_selection import train_test_split

# authentication
# If you have not previously saved your credentials, follow instructions at
# /docs/guides/functions
# to authenticate with your API key.
catalog = QiskitFunctionsCatalog(channel="ibm_quantum_platform")

# load "Singularity Machine Learning - Classification" function by Multiverse Computing
singularity = catalog.load("multiverse/singularity")

# generate the synthetic dataset
X, y = make_moons(n_samples=1000)

# split the data into training and test datasets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)

job = singularity.run(
    action="create_fit_predict",
    num_learners=10,
    regularization=0.01,
    optimizer_options={"simulator": True},
    X_train=X_train,
    y_train=y_train,
    X_test=X_test,
    options={"save": False},
)

# get job status and result
status = job.status()
result = job.result()

print("Job status: ", status)
print("Action result status: ", result["status"])
print("Action result message: ", result["message"])
print("Predictions (first five results): ", result["data"]["predictions"][:5])
print(
    "Probabilities (first five results): ",
    result["data"]["probabilities"][:5],
)
print("Usage metadata: ", result["metadata"]["resource_usage"])

Output:

Job status:  QUEUED
Action result status:  ok
Action result message:  Classifier created, fitted, and predicted.
Predictions (first five results):  [0, 0, 1, 0, 0]
Probabilities (first five results):  [[0.87119766531518, 0.1288023346848197], [0.87119766531518, 0.1288023346848197], [0.24470328446479797, 0.7552967155352032], [0.820524432250189, 0.17947556774981072], [0.6847610293419495, 0.31523897065805173]]
Usage metadata:  {'RUNNING: MAPPING': {'CPU_TIME': 10.967791318893433}, 'RUNNING: WAITING_QPU': {'CPU_TIME': 59.91712307929993}, 'RUNNING: POST_PROCESSING': {'CPU_TIME': 59.097386837005615}, 'RUNNING: EXECUTING_QPU': {'QPU_TIME': 56.93338203430176}}

Get started with Singularity Machine Learning

Set up

Examples

Classify a dataset

create_fit_predict example

`create_fit_predict` example