232 research outputs found
Robustness of quantum reinforcement learning under hardware errors
No full textVariational quantum machine learning algorithms have become the focus of recent research on how to utilize near-term quantum devices for machine learning tasks. They are considered suitable for this as the circuits that are run can be tailored to the device, and a big part of the computation is delegated to the classical optimizer. It has also been hypothesized that they may be more robust to hardware noise than conventional algorithms due to their hybrid nature. However, the effect of training quantum machine learning models under the influence of hardware-induced noise has not yet been extensively studied. In this work, we address this question for a specific type of learning, namely variational reinforcement learning, by studying its performance in the presence of various noise sources: shot noise, coherent and incoherent errors. We analytically and empirically investigate how the presence of noise during training and evaluation of variational quantum reinforcement learning algorithms affect the performance of the agents and robustness of the learned policies. Furthermore, we provide a method to reduce the number of measurements required to train Q-learning agents, using the inherent structure of the algorithm
Quantum Policy Gradient Algorithms
No full textUnderstanding the power and limitations of quantum access to data in machine learning tasks is primordial to assess the potential of quantum computing in artificial intelligence. Previous works have already shown that speed-ups in learning are possible when given quantum access to reinforcement learning environments. Yet, the applicability of quantum algorithms in this setting remains very limited, notably in environments with large state and action spaces. In this work, we design quantum algorithms to train state-of-the-art reinforcement learning policies by exploiting quantum interactions with an environment. However, these algorithms only offer full quadratic speed-ups in sample complexity over their classical analogs when the trained policies satisfy some regularity conditions. Interestingly, we find that reinforcement learning policies derived from parametrized quantum circuits are well-behaved with respect to these conditions, which showcases the benefit of a fully-quantum reinforcement learning framework
Veb pristupačnost i elektronskih formati za pristup informacijama za osobe sa invaliditetom
No full textAuthor described the context of the development of technological and social relationships and how they mutually made changes on ways how people interact with information and access to knowledge.
The development of societies changed significantly the role of libraries and posed new technological and other challenges due to the development of information technologies.
Since the development of information technologies caused expansion of production capacities in societies there was additional need to standardize production of formats of access, distribution, archiving of information. Due to democratic character of many developed societies openness of information and human rights emphasized an importance of legal aspects and rights of persons with disabilities. Author mentioned articles of the UN Convention of Rights of Persons with Disabilities which required that parties which ratified convention should implement accessibility standards for persons with disabilities. Author presented standards and technical specifications that define accessibility of web interface, computer software and file formats. It is especially emphasized that those standards are important in libraries. Modern libraries should adopt and use file formats which are accessible for persons with disabilities in order to avoid risk of increased discrimination against persons with disabilities if inaccessible technologies will be used. Author believes that this would help persons with disability to use library resources and services equally as other users do.
Ideal quantum protocols in the non-ideal physical world
No full textThe development of quantum protocols from conception to experimental realizations is one of
the main sources of the stimulating exchange between fundamental and experimental research
characteristic to quantum information processing. In this thesis we contribute to the development
of two recent quantum protocols, Universal Blind Quantum Computation (UBQC) and Quantum
Digital Signatures (QDS). UBQC allows a client to delegate a quantum computation to a more
powerful quantum server while keeping the input and computation private. We analyse the resilience
of the privacy of UBQC under imperfections. Then, we introduce approximate blindness
quantifying any compromise to privacy, and propose a protocol which enables arbitrary levels of
security despite imperfections. Subsequently, we investigate the adaptability of UBQC to alternative
implementations with practical advantages. QDS allow a party to send a message to other
parties which cannot be forged, modified or repudiated. We analyse the security properties of a
first proof-of-principle experiment of QDS, implemented in an optical system. We estimate the
security failure probabilities of our system as a function of protocol parameters, under all but the
most general types of attacks. Additionally, we develop new techniques for analysing transformations
between symmetric sets of states, utilized not only in the security proofs of QDS but in
other applications as well
Does strict employment protection discourage job creation? Evidence from Croatia
No full textEmployment protection legislation in Croatia is among the most strict in Europe. Firing is difficult and costly, and flexible forms of employment are limited. Is this apparent rigidity reflected-as one would expect based on standard economic theory-in low labor market dynamics? Is job creation low and hiring limited? Is the job security of insiders achieved at the cost of outsiders not being able to enter thelabor market? The author attempts to answer these questions by examining job flows. If the employment protection legislation is binding, then job and worker turnover should be low. He shows that this is indeed the case. Hiring is limited and the average job tenure is very long in Croatia. Job destruction is low, however job creation is still lower. The result is accumulation of unemployment, in large part due to new labor market entrants not being able to find a job. The high degree of job protection also seems to strengthen the bargaining position of insiders and results in relatively high wages. So, wages in Croatia are higher than among its competitors, even after adjusting for productivity. These high labor costs are likely to contribute to limited job creation in existing firms, but also are likely to discourage the entry of-and thus job creation in-new firms. The author presents evidence that firm growth has been indeed limited in Croatia, contributing to the low employment level. The author examines other potential causes of high unemployment in Croatia (the unemployment benefit system, labor taxation, the wage structure, and skill and spatial mismatches). He argues that they do not play a substantial part in accounting for poor labor market outcomes in Croatia. The author concludes that the stringent employment protection legislation is the key labor market institution behind low job creation and high unemployment. Based on this he recommends specific measures aimed at liberalizing the labor market to foster job creation and employment.Labor Management and Relations,Labor Policies,Labor Markets,Environmental Economics&Policies,Trade Finance and Investment,Labor Markets,Labor Management and Relations,Labor Standards,Banks&Banking Reform,Environmental Economics&Policies
Quantum Techniques in Machine Learning (QTML conference 2023)
No full textAbstract:
Although still a relatively niche field in classical machine learning, topological data analysis has raised substantial interest from the perspective of quantum algorithms in the last few years.
In this talk we will introduce the topic of topological data analysis, and discuss the state-of-art of quantum algorithms for this problem, together with their promises and limitations, possible generalisations and connections to many-body physics
Truly noiseless probabilistic amplification
No full textMost of the schemes for "noiseless" amplification of coherent states, which have recently been attracting theoretical and experimental interest, share a common trait: The amplification is not truly noiseless, or perfect, for nonzero success probability. While this must hold true for all phase-independent amplification schemes, in this work we point out that truly noiseless amplification is indeed possible, provided that the states which we wish to amplify come from a finite set. Perfect amplification with unlimited average gain is then possible with finite success probability, for example, using techniques for unambiguously distinguishing between quantum states. Such realizations require only linear optics, no single-photon sources, and no photon counting. We also investigate the optimal success probability of perfect amplification of a symmetric set of coherent states. There are two regimes: low-amplitude amplification, where the target amplitude is below one, and general amplification. For the low-amplitude regime, analytic results for the optimal amplification success probabilities can be obtained. In this case a natural bound imposed by the ratio of success probabilities of optimal unambiguous discrimination of the source and amplified states can always be reached. We also show that for general amplification this bound cannot always be satisfied.</p
Extended phase map decompositions for unitaries
No full textWe give a complete structural characterisation of the map implemented by the positive branch of a one-way pattern. Our approach is based on the phase map decomposition (de Beaudrap et al. 2006; de Beaudrap et al. 2008) and leads to some preliminary results on the connection between the column structure of a given unitary and the angles of measurements in a pattern that implements it. Our characterisation highlights the role of entanglement in the efficiency of the simulation of a one-way pattern, and it is a step forward towards a full characterisation of those unitaries that have an efficient one-way model implementation.</p
- …
