4.3 Classical Benchmarks
Classical benchmarking is an important element of the testing of quantum algorithms. Small-scale (or even stylised) problems are ideally suited for this task. Let us see how the QBoost model performs in comparison with the standard classical ML classifiers: neural networks and gradient boosting.
4.3.1 Artificial neural network
An Artificial Neural Network (ANN) is a network of interconnected activation units (or artificial neurons), where each activation unit performs three main functions (Figure 4.3):
- Summation of the input signals (xi)i=1,…,N, from all the upstream units to which it is connected with multiplication by the corresponding weights (wi)i=1,…,N;
- Non-linear transformation of the aggregated input;
- Sending the result to the downstream units to which it is connected.
Sometimes the activation unit also performs binarisation (or, more generally, digitisation) of the output – typically, this is a task of the activation...
