The issue can be overcome by the second proposed algorithm, which includes a quantum random access memory, amplitude estimation, and the Dürr Høyer minimization algorithm. However, this is inefficient in few situations like when the centroid of the subset lies in the space of other subsets. First is the quantum version of the nearest centroid algorithm which shows an exponential speedup for both the size of the training set and the dimension of the vectors. Recently proposed binary quantum decision tree classifier, named Q-tree, is based on a probabilistic approach where a quantum computer can be used to accomplish the probabilistic traversal of the decision tree via measurements, while tree inductions and predictions of query data can be integrated into this framework as well.įor the case of the quantum nearest neighbors algorithm, the authors discuss two algorithms. In that scenario, the quantum computer can be used to map the classical data to a Hilbert state space and estimate the inner products between those states to obtain the kernel matrix, which is further handled by classical computers.Īnother strong candidate for classification is the quantum decision tree model, that uses von Neumann entropy for choosing the attributes to split the nodes. So far, quantum kernel methods used in quantum support vector machines are proven to be the most successful to handle classification tasks. The experimental advancements of the considered classifiers are also discussed along with theoretical progress. This review paper introduces some important algorithms and models for quantum classifiers including i) quantum support vector machines, ii) quantum decision tree, iii) quantum nearest neighbors algorithms, iv) quantum annealing based classifiers, and v) variational quantum classifiers. Previous studies on quantum classifiers have extended popular classical classification algorithms to the quantum domain such as the quantum support vector machine, quantum nearest neighbor algorithms, and quantum decision tree classifiers, exhibiting the potential of providing quadratic or even exponential speedups. However, one of the most well-known branches of machine learning is classification, which is widely applied in commercial and academic applications ranging from face recognition and recommendation systems to earthquake detection, disease diagnosis and further handling complex classification tasks. Some notable examples are image recognition advancements, automated driven cars, identification of skin cancers, and prediction of protein structures. Recently, there has been extensive research into quantum machine learning models and quantum algorithms that rival their classical counterparts.
0 Comments
Leave a Reply. |