TNO Quantum: Machine Learning - Regression - Linear regression

TNO Quantum provides generic software components aimed at facilitating the development of quantum applications.

This package contains implementations of quantum-inspired algorithms for linear regression. It assumes a linear system of the form Ax=b, where A is the training data, x is a vector of unknown coefficients, and b is a vector of target values.

The class QILinearEstimator provides three methods: fit, predict_x, and predict_b. Once the fit method has been called using A and b, predict_x can be used to sample entries of the estimated coefficient vector. Alternatively, predict_b can be used to sample entries of predictions corresponding to (un)observed target values.

Limitations in (end-)use: the content of this software package may solely be used for applications that comply with international export control laws.

Documentation

Documentation of the linear regression package can be found [here](https://tno-
quantum.github.io/documentation/).

Install

Easily install the tno.quantum.ml.regression.linear_regression package using pip:

$ python -m pip install tno.quantum.ml.regression.linear_regression

Example

See example below. More examples can be found in the test directory.

import numpy as np
from sklearn.datasets import make_low_rank_matrix
from sklearn.model_selection import train_test_split
from tno.quantum.ml.regression.linear_regression import QILinearEstimator

rng = np.random.RandomState(7)

# Generate example data
m = 700
n = 100
A = make_low_rank_matrix(n_samples=m, n_features=n, effective_rank=3, random_state=rng, tail_strength=0.1)
x = rng.normal(0, 1, A.shape[1])
b = A @ x

# Create training and test datasets
A_train, A_test, b_train, b_test = train_test_split(A, b, test_size=0.3, random_state=rng)

# Fit quantum-inspired model
rank = 3
r = 100
c = 30
n_samples = 100  # for Monte Carlo methods
qi = QILinearEstimator(r, c, rank, n_samples, rng, sketcher_name="fkv")
qi = qi.fit(A_train, b_train)

# Sample from b (vector of predictions)
n_entries_b = 1000
sampled_indices_b, sampled_b = qi.predict_b(A_test, n_entries_b)

Credits

The algorithms found in this repository have been developed in collaboration with the Quantum Application Lab and have been based on:

• https://github.com/QuantumApplicationLab/quantum-inspired-algorithms
• https://github.com/XanaduAI/quantum-inspired-algorithms
• "Quantum-inspired algorithms in practice", by Juan Miguel Arrazola, Alain Delgado, Bhaskar Roy Bardhan, and Seth Lloyd. 2020-08-13, volume 4, page 307. Quantum 4, 307 (2020).
• "Quantum-inspired low-rank stochastic regression with logarithmic dependence on the dimension", by András Gilyén, Seth Lloyd, Ewin Tang. (2018). ArXiv, abs/1811.04909.

This work was supported by the Dutch National Growth Fund (NGF), as part of the Quantum Delta NL programme.