| Quantum artificial intelligence is a combination of quantum mechanics and artifi-cial intelligence technology.A quantum system has coherence,that is to say,the system can have the characteristics of multiple state combinations at the same time.This char-acteristic makes quantum system have the advantage of inherent parallel computing when it is used as computing bit.The derived quantum computing science,which is to build a quantum computer composed of quantum bits to greatly improve the computing power,is a hot research direction at present.At the same time,the rapid development of artificial intelligence technology in recent years has gradually become an integral part of our daily life,such as image recognition,automatic driving,urban brain and so on.With the advent of the era of big data,the requirements of artificial intelligence for computer computing power are also constantly improving,which makes classic computers con-stantly approach the bottleneck restricted by Moore's law.Under such a background,we want to integrate quantum mechanics,especially quantum computing,with artificial intelligence,and use the computing power advantage brought by quantum parallel com-puting to serve more complex and large amount of data artificial intelligence algorithm.There are many physical systems for quantum computing,such as superconducting system,ion trap system and so on.Linear optical systems usually take the polarization and path of photon state as quantum bits,and control the whole state of each quantum bit to achieve the corresponding computing task.This system has many advantages,such as long coherent time,many coding dimensions,and can run at room temperature.It can support the quantum version of a variety of artificial intelligence algorithms and shorten the corresponding operation time.The followings are the main research works of my doctoral dissertation:1.Experimental study on searching the optimal quantum coherent freezing point by gradient descent methodQuantum coherence is an important resource of quantum system.When passing through a noisy channel,the quantum coherence will decrease with the increase of evo-lution time.We try to use wave plate to control various parameters of quantum channel,and use the most primitive gradient descent algorithm to search the channel state which can keep the maximum coherence after a period of evolution time.It is found that the quantum channel after learning can make the quantum coherence resource to minimize or even freeze during evolution.2.Detection of quantum mutation points by Bayesian inferenceBecause of the coherence of quantum system,we can only distinguish two non orthogonal quantum states with certain probability.When a series of quantum states are emitted continuously and change at some point,how to detect the location of the mutation with the highest success rate is the problem we need to solve.In this paper,we use Bayesian inference to measure every quantum state continuously,and use Bayesian update to adjust the measurement basis quickly after each measurement.Finally,we can find the mutation point of quantum state with the highest accuracy through the analysis of prior probability.3.Reconstruction of single quantum states by deep learning approachQuantum state reconstruction is an important part of quantum information research,and the most commonly used method is quantum tomography.However,this method generally needs a large number of quantum state copies as the guarantee of accuracy.Using the deep learning algorithm,we can adjust the measurement basis according to the measured data each time.We find that we can use fewer copies of quantum states to complete the same precision quantum state reconstruction task.4.Using machine learning method to predict the dynamic process of the sys-tem under non-Markovian evolutionBecause the quantum system is always inevitably coupled with the environment,the quantum open system will inevitably be involved in the study of various practical problems.For an unknown quantum channel with memory,we use machine learning and Riemann optimization to predict the evolution of a quantum system. |
PDF Full Text Request