Machine learning algorithms learn a desired input-output relation from examples in order to interpret new inputs. This is important for tasks such as image and speech recognition or strategy optimisation, with growing applications in the IT industry. In the last couple of years, researchers investigated if quantum computing can help to improve classical machine learning algorithms. Ideas range from running computationally costly algorithms or their subroutines efficiently on a quantum computer to the translation of stochastic methods into the language of quantum theory. This contribution gives a systematic overview of the emerging field of quantum machine learning. It presents the approaches as well as technical details in an accessable way, and discusses the potential of a future theory of quantum learning.
With the overwhelming success in the field of quantum information in the last decades, the quest for a Quantum Neural Network (QNN) model began in order to combine quantum computing with the striking properties of neural computing. This article presents a systematic approach to QNN research, which so far consists of a conglomeration of ideas and proposals. It outlines the challenge of combining the nonlinear, dissipative dynamics of neural computing and the linear, unitary dynamics of quantum computing. It establishes requirements for a meaningful QNN and reviews existing literature against these requirements. It is found that none of the proposals for a potential QNN model fully exploits both the advantages of quantum physics and computing in neural networks. An outlook on possible ways forward is given, emphasizing the idea of Open Quantum Neural Networks based on dissipative quantum computing.
A quantum algorithm is presented for the simulation of arbitrary Markovian dynamics of a qubit, described by a semigroup of single qubit quantum channels ${T_t}$ specified by a generator $mathcal{L}$. This algorithm requires only $mathcal{O}big((||mathcal{L}||_{(1rightarrow 1)} t)^{3/2}/epsilon^{1/2} big)$ single qubit and CNOT gates and approximates the channel $T_t = e^{tmathcal{L}}$ up to chosen accuracy $epsilon$. Inspired by developments in Hamiltonian simulation, a decomposition and recombination technique is utilised which allows for the exploitation of recently developed methods for the approximation of arbitrary single-qubit channels. In particular, as a result of these methods the algorithm requires only a single ancilla qubit, the minimal possible dilation for a non-unitary single-qubit quantum channel.
A connection between the asymptotic behavior of the open quantum walk and the spectrum of a generalized quantum coins is studied. For the case of simultaneously diagonalizable transition operators an exact expression for probability distribution of the position of the walker for arbitrary number of steps is found. For a large number of steps the probability distribution consist of maximally two soliton-like solution and a certain number of Gaussian distributions. The number of different contributions to the final probability distribution is equal to the number of distinct absolute values in the spectrum of the transition operators. The presence of the zeros in spectrum is an indicator of the soliton-like solutions in the probability distribution.
A universal definition of non-Markovianity for open systems dynamics is proposed. It is extended from the classical definition to the quantum realm by showing that a `transition from the Markov to the non-Markov regime occurs when the correlations between the system and the environment, generated by their joint evolution, can no longer be neglected. The suggested definition is based on the comparison between measured correlation functions and those built by assuming that the system is in a Markov regime thus giving a figure of merit of the error coming from this assumption. It is shown that the knowledge of the dynamical map and initial condition of the system is not enough to fully characterise the non-Markovian dynamics of the reduced system. The example of three exactly solvable models, i.e. decoherence and spontaneous emission of the qubit in a bosonic bath and decoherence of the photons polarization induced by interaction with its spacial degrees of freedom, reveals that previously proposed Markovianity criteria and measures which are based on dynamical map analysis fail to recognise non-Markov behaviour.
The dynamics of two interacting spins coupled to separate bosonic baths is studied. An analytical solution in Born approximation for arbitrary spectral density functions of the bosonic environments is found. It is shown that in the non-Markovian cases concurrence lives longer or reaches greater values.
