How to call EnergyInput() function in Qiskit python?

Question

I want to implement SVM with Qiskit. I used this following code.

from qiskit import Aer
from qiskit.aqua.utils import split_dataset_to_data_and_labels
from qiskit.aqua.input import get_input_instance     
from qiskit.aqua import run_algorithm

n = 2  # dimension of each data point
sample_Total, training_input, test_input, class_labels = Breast_cancer(training_size=40,
                                                              test_size=10, n=n, PLOT_DATA=True)

temp = [test_input[k] for k in test_input]
total_array = np.concatenate(temp)

aqua_dict = {
    'problem': {'name': 'svm_classification', 'random_seed': 100},
    'algorithm': {
        'name': 'QSVM.Kernel'
    },
    'feature_map': {'name': 'SecondOrderExpansion', 'depth': 2, 'entangler_map': {0: [1]}},
    'multiclass_extension': {'name': 'AllPairs'},
    'backend': {'name': 'qasm_simulator', 'shots': 256}
}

algo_input = get_input_instance('SVMInput')
algo_input.training_dataset = training_input
algo_input.test_dataset = test_input
algo_input.datapoints = total_array

result = run_algorithm(aqua_dict, algo_input)
for k,v in result.items():
    print("'{}' : {}".format(k, v))

But this code shows this error

ImportError: cannot import name 'get_input_instance'

This is because this method is removed from Qiskit. I got this piece of information from this github issue. They have suggested to use EnergyInput() instead of get_input_instance() in a similar way. So I modified the previous code in the following way.

!pip install qiskit
from qiskit import Aer
from qiskit.aqua.utils import split_dataset_to_data_and_labels
from qiskit.aqua.input import EnergyInput
from qiskit.aqua import run_algorithm

algo_input = EnergyInput('SVMInput')
algo_input.training_dataset = training_input
algo_input.test_dataset = test_input
algo_input.datapoints = total_array

Now this code shows that EnergyInput cannot take anykind of String input. This generates the following error.

AttributeError: 'str' object has no attribute 'to_dict'

I can show the whole error in this screenshot.

Answer 1

Have a look at this tutorial about creating QSVMs. Instead of EnergyInput() they use a class called ClassificationInput() to which they pass their data.

This makes the overall expression: algo_input = ClassificationInput(training_input, test_input, datapoints[0])