Binary Blobs

Data for benchmarking machine learning models, taken from Train on classical, deploy on quantum: scaling generative quantum machine learning to a thousand qubits. The Binary Blobs dataset can be seen as a binary version of the Gaussian blobs dataset for continuous data. This dataset is the specific dataset that appeared in the above paper; more general datasets can be constructed via the qml_benchmarks package.

Description of the dataset

The dataset consists of bit strings of length 16. To generate samples, one of the 8 following patterns is selected at random (where data has been reshaped to size (4,4))

Each bit is then flipped with 5% probability. There are 5000 training points and 10000 test points. If needed, labels that correspond to the 8 patterns can also be accessed. Please see the Source code tab to check how the data was generated.

Example usage

>>> [ds] = qml.data.load("other", name="binary-blobs")
>>>
>>> blob_vector = ds.train['inputs'][0]
>>> blob_array = np.reshape(blob_vector, (4,4))
>>> print(blob_array)
[[0. 0. 1. 1.]
 [0. 0. 1. 1.]
 [0. 0. 0. 0.]
 [0. 0. 0. 0.]]
>>> ds.train['labels'][0]
1
>>> blob_vector = ds.test('inputs')[10]
>>> blob_array = np.reshape(blob_vector, (4,4))
>>> print(blob_array)
[[1. 0. 1. 0.]
 [0. 1. 0. 0.]
 [0. 0. 1. 0.]
 [0. 0. 0. 1.]]
>>> ds.test['labels'][10]
5

version 0.1 : initial public release