arXiv subjects

Yuxiang Yang

Publications and source records attributed to Yuxiang Yang.

At least 109 records · Page 6Linked to original sources

arXiv2020

Rapidly Adaptable Legged Robots via Evolutionary Meta-Learning

Learning adaptable policies is crucial for robots to operate autonomously in our complex and quickly changing world. In this work, we present a new meta-learning method that allows robots to quickly adapt to changes in dynamics. In contrast to gradient-based meta-learning algorithms that rely on second-order gradient estimation, we introduce a more noise-tolerant Batch Hill-Climbing adaptation operator and combine it with meta-learning based on evolutionary strategies. Our method significantly improves adaptation to changes in dynamics in high noise settings, which are common in robotics applications. We validate our approach on a quadruped robot that learns to walk while subject to changes in dynamics. We observe that our method significantly outperforms prior gradient-based approaches, enabling the robot to adapt its policy to changes based on less than 3 minutes of real data.

cs.RO↗

arXiv2020

ES-MAML: Simple Hessian-Free Meta Learning

We introduce ES-MAML, a new framework for solving the model agnostic meta learning (MAML) problem based on Evolution Strategies (ES). Existing algorithms for MAML are based on policy gradients, and incur significant difficulties when attempting to estimate second derivatives using backpropagation on stochastic policies. We show how ES can be applied to MAML to obtain an algorithm which avoids the problem of estimating second derivatives, and is also conceptually simple and easy to implement. Moreover, ES-MAML can handle new types of nonsmooth adaptation operators, and other techniques for improving performance and estimation of ES methods become applicable. We show empirically that ES-MAML is competitive with existing methods and often yields better adaptation with fewer queries.

cs.LG↗

arXiv2020

Quantum state estimation with nuisance parameters

In parameter estimation, nuisance parameters refer to parameters that are not of interest but nevertheless affect the precision of estimating other parameters of interest. For instance, the strength of noises in a probe can be regarded as a nuisance parameter. Despite its long history in classical statistics, the nuisance parameter problem in quantum estimation remains largely unexplored. The goal of this article is to provide a systematic review of quantum estimation in the presence of nuisance parameters, and to supply those who work in quantum tomography and quantum metrology with tools to tackle relevant problems. After an introduction to the nuisance parameter and quantum estimation theory, we explicitly formulate the problem of quantum state estimation with nuisance parameters. We extend quantum Cramer-Rao bounds to the nuisance parameter case and provide a parameter orthogonalization tool to separate the nuisance parameters from the parameters of interest. In particular, we put more focus on the case of one-parameter estimation in the presence of nuisance parameters, as it is most frequently encountered in practice.

quant-ph↗

arXiv2020

Quantum Metrology with Indefinite Causal Order

We address the study of quantum metrology enhanced by indefinite causal order, demonstrating a quadratic advantage in the estimation of the product of two average displacements in a continuous variable system. We prove that no setup where the displacements are probed in a fixed order can have root-mean-square error vanishing faster than the Heisenberg limit 1/N, where N is the number of displacements contributing to the average. In stark contrast, we show that a setup that probes the displacements in a superposition of two alternative orders yields a root-mean-square error vanishing with super-Heisenberg scaling 1/N^2. This result opens up the study of new measurement setups where quantum processes are probed in an indefinite order, and suggests enhanced tests of the canonical commutation relations, with potential applications to quantum gravity.

quant-ph↗

arXiv2020

Ultimate limit on time signal generation

The generation of time signals is a fundamental task in science. Here we study the relation between the quality of a time signal and the physics of the system that generates it. According to quantum theory, any time signal can be decomposed into individual quanta that lead to single detection events. Our main result is a bound on how sharply peaked in time these events can be, which depends on the dimension of the signal generator. This result promises applications in various directions, including information theory, quantum clocks, and process simulation.

quant-ph↗

arXiv2020

Entropic time-energy uncertainty relations: An algebraic approach

We address entropic uncertainty relations between time and energy or, more precisely, between measurements of an observable $G$ and the displacement $r$ of the $G$-generated evolution $e^{-ir G}$. We derive lower bounds on the entropic uncertainty in two frequently considered scenarios, which can be illustrated as two different guessing games in which the role of the guessers are fixed or not. In particular, our bound for the first game improves the previous result by Coles et al.. Our derivation uses as a subroutine a recently proposed novel algebraic method, which can in general be used to derive a wider class of entropic uncertainty principles.

quant-ph↗

arXiv2019

Deep Time-Frequency Representation and Progressive Decision Fusion for ECG Classification

Early recognition of abnormal rhythms in ECG signals is crucial for monitoring and diagnosing patients' cardiac conditions, increasing the success rate of the treatment. Classifying abnormal rhythms into exact categories is very challenging due to the broad taxonomy of rhythms, noises and lack of large-scale real-world annotated data. Different from previous methods that utilize hand-crafted features or learn features from the original signal domain, we propose a novel ECG classification method by learning deep time-frequency representation and progressive decision fusion at different temporal scales in an end-to-end manner. First, the ECG wave signal is transformed into the time-frequency domain by using the Short-Time Fourier Transform. Next, several scale-specific deep convolutional neural networks are trained on ECG samples of a specific length. Finally, a progressive online decision fusion method is proposed to fuse decisions from the scale-specific models into a more accurate and stable one. Extensive experiments on both synthetic and real-world ECG datasets demonstrate the effectiveness and efficiency of the proposed method.

cs.LG↗

arXiv2019

Accuracy enhancing protocols for quantum clocks

The accuracy of the time information generated by clocks can be enhanced by allowing them to communicate with each other. Here we consider a basic scenario where a quantum clock receives a low-accuracy time signal as input and ask whether it can generate an output of higher accuracy. We propose protocols that use a quantum clock with a $d$-dimensional state space to achieve an accuracy enhancement by a factor of $d$, for large enough $d$. If no feedback to the input signal is allowed, this enhancement is temporary. With feedback the accuracy enhancement can be retained indefinitely. Our protocols are specific to quantum clocks, and may be used to synchronise them in a network, defining a time scale that is more accurate than what can be achieved by non-interacting or classical clocks.

quant-ph↗

arXiv2019

A unified framework of predicting binary interestingness of images based on discriminant correlation analysis and multiple kernel learning

In the modern content-based image retrieval systems, there is an increasingly interest in constructing a computationally effective model to predict the interestingness of images since the measure of image interestingness could improve the human-centered search satisfaction and the user experience in different applications. In this paper, we propose a unified framework to predict the binary interestingness of images based on discriminant correlation analysis (DCA) and multiple kernel learning (MKL) techniques. More specially, on the one hand, to reduce feature redundancy in describing the interestingness cues of images, the DCA or multi-set discriminant correlation analysis (MDCA) technique is adopted to fuse multiple feature sets of the same type for individual cues by taking into account the class structure among the samples involved to describe the three classical interestingness cues, unusualness,aesthetics as well as general preferences, with three sets of compact and representative features; on the other hand, to make good use of the heterogeneity from the three sets of high-level features for describing the interestingness cues, the SimpleMKL method is employed to enhance the generalization ability of the built model for the task of the binary interestingness classification. Experimental results on the publicly-released interestingness prediction data set have demonstrated the rationality and effectiveness of the proposed framework in the binary prediction of image interestingness where we have conducted several groups of comparative studies across different interestingness feature combinations, different interestingness cues, as well as different feature types for the three interestingness cues.

cs.CV↗

arXiv2019

Data Efficient Reinforcement Learning for Legged Robots

We present a model-based framework for robot locomotion that achieves walking based on only 4.5 minutes (45,000 control steps) of data collected on a quadruped robot. To accurately model the robot's dynamics over a long horizon, we introduce a loss function that tracks the model's prediction over multiple timesteps. We adapt model predictive control to account for planning latency, which allows the learned model to be used for real time control. Additionally, to ensure safe exploration during model learning, we embed prior knowledge of leg trajectories into the action space. The resulting system achieves fast and robust locomotion. Unlike model-free methods, which optimize for a particular task, our planner can use the same learned dynamics for various tasks, simply by changing the reward function. To the best of our knowledge, our approach is more than an order of magnitude more sample efficient than current model-free methods.

cs.LG↗

arXiv2019

Memory effects in quantum metrology

Quantum metrology concerns estimating a parameter from multiple identical uses of a quantum channel. We extend quantum metrology beyond this standard setting and consider estimation of a physical process with quantum memory, here referred to as a parametrized quantum comb. We present a theoretic framework of metrology of quantum combs, and derive a general upper bound of the comb quantum Fisher information. The bound can be operationally interpreted as the quantum Fisher information of a memoryless quantum channel times a dimensional factor. We then show an example where the bound can be attained up to a factor of four. With the example and the bound, we show that memory in quantum sensors plays an even more crucial role in the estimation of combs than in the standard setting of quantum metrology.

quant-ph↗

arXiv2019

Provably Robust Blackbox Optimization for Reinforcement Learning

Interest in derivative-free optimization (DFO) and "evolutionary strategies" (ES) has recently surged in the Reinforcement Learning (RL) community, with growing evidence that they can match state of the art methods for policy optimization problems in Robotics. However, it is well known that DFO methods suffer from prohibitively high sampling complexity. They can also be very sensitive to noisy rewards and stochastic dynamics. In this paper, we propose a new class of algorithms, called Robust Blackbox Optimization (RBO). Remarkably, even if up to $23\%$ of all the measurements are arbitrarily corrupted, RBO can provably recover gradients to high accuracy. RBO relies on learning gradient flows using robust regression methods to enable off-policy updates. On several MuJoCo robot control tasks, when all other RL approaches collapse in the presence of adversarial noise, RBO is able to train policies effectively. We also show that RBO can be applied to legged locomotion tasks including path tracking for quadruped robots.

cs.LG↗

arXiv2019

Attaining the ultimate precision limit in quantum state estimation

We derive a bound on the precision of state estimation for finite dimensional quantum systems and prove its attainability in the generic case where the spectrum is non-degenerate. Our results hold under an assumption called local asymptotic covariance, which is weaker than unbiasedness or local unbiasedness. The derivation is based on an analysis of the limiting distribution of the estimator's deviation from the true value of the parameter, and takes advantage of quantum local asymptotic normality, a useful asymptotic characterization of identically prepared states in terms of Gaussian states. We first prove our results for the mean square error of a special class of models, called D-invariant, and then extend the results to arbitrary models, generic cost functions, and global state estimation, where the unknown parameter is not restricted to a local neighbourhood of the true value. The extension includes a treatment of nuisance parameters, i.e. parameters that are not of interest to the experimenter but nevertheless affect the precision of the estimation. As an illustration of the general approach, we provide the optimal estimation strategies for the joint measurement of two qubit observables, for the estimation of qubit states in the presence of amplitude damping noise, and for noisy multiphase estimation.

quant-ph↗

arXiv2019

NoRML: No-Reward Meta Learning

Efficiently adapting to new environments and changes in dynamics is critical for agents to successfully operate in the real world. Reinforcement learning (RL) based approaches typically rely on external reward feedback for adaptation. However, in many scenarios this reward signal might not be readily available for the target task, or the difference between the environments can be implicit and only observable from the dynamics. To this end, we introduce a method that allows for self-adaptation of learned policies: No-Reward Meta Learning (NoRML). NoRML extends Model Agnostic Meta Learning (MAML) for RL and uses observable dynamics of the environment instead of an explicit reward function in MAML's finetune step. Our method has a more expressive update step than MAML, while maintaining MAML's gradient based foundation. Additionally, in order to allow more targeted exploration, we implement an extension to MAML that effectively disconnects the meta-policy parameters from the fine-tuned policies' parameters. We first study our method on a number of synthetic control problems and then validate our method on common benchmark environments, showing that NoRML outperforms MAML when the dynamics change between tasks.

cs.LG↗

arXiv2019

OpenRoACH: A Durable Open-Source Hexapedal Platform with Onboard Robot Operating System (ROS)

OpenRoACH is a 15-cm 200-gram self-contained hexapedal robot with an onboard single-board computer. To our knowledge, it is the smallest legged robot with the capability of running the Robot Operating System (ROS) onboard. The robot is fully open sourced, uses accessible materials and off-the-shelf electronic components, can be fabricated with benchtop fast-prototyping machines such as a laser cutter and a 3D printer, and can be assembled by one person within two hours. Its sensory capacity has been tested with gyroscopes, accelerometers, Beacon sensors, color vision sensors, linescan sensors and cameras. It is low-cost within \$150 including structure materials, motors, electronics, and a battery. The capabilities of OpenRoACH are demonstrated with multi-surface walking and running, 24-hour continuous walking burn-ins, carrying 200-gram dynamic payloads and 800-gram static payloads, and ROS control of steering based on camera feedback. Information and files related to mechanical design, fabrication, assembly, electronics, and control algorithms are all publicly available on https://wiki.eecs.berkeley.edu/biomimetics/Main/OpenRoACH.

cs.RO↗

arXiv2019

Quantum Stopwatch: How To Store Time in a Quantum Memory

Quantum mechanics imposes a fundamental tradeoff between the accuracy of time measurements and the size of the systems used as clocks. When the measurements of different time intervals are combined, the errors due to the finite clock size accumulate, resulting in an overall inaccuracy that grows with the complexity of the setup. Here we introduce a method that in principle eludes the accumulation of errors by coherently transferring information from a quantum clock to a quantum memory of the smallest possible size. Our method could be used to measure the total duration of a sequence of events with enhanced accuracy, and to reduce the amount of quantum communication needed to stabilize clocks in a quantum network.

quant-ph↗

arXiv2019

Compression for quantum population coding

We study the compression of n quantum systems, each prepared in the same state belonging to a given parametric family of quantum states. For a family of states with f independent parameters, we devise an asymptotically faithful protocol that requires a hybrid memory of size (f/2)log(n), including both quantum and classical bits. Our construction uses a quantum version of local asymptotic normality and, as an intermediate step, solves the problem of compressing displaced thermal states of n identically prepared modes. In both cases, we show that (f/2)log(n) is the minimum amount of memory needed to achieve asymptotic faithfulness. In addition, we analyze how much of the memory needs to be quantum. We find that the ratio between quantum and classical bits can be made arbitrarily small, but cannot reach zero: unless all the quantum states in the family commute, no protocol using only classical bits can be faithful, even if it uses an arbitrarily large number of classical bits.

quant-ph↗

arXiv2017

Units of rotational information

Entanglement in angular momentum degrees of freedom is a precious resource for quantum metrology and control. Here we study the conversions of this resource, focusing on Bell pairs of spin-J particles, where one particle is used to probe unknown rotations and the other particle is used as reference. When a large number of pairs are given, we show that every rotated spin-J Bell state can be reversibly converted into an equivalent number of rotated spin one-half Bell states, at a rate determined by the quantum Fisher information. This result provides the foundation for the definition of an elementary unit of information about rotations in space, which we call the Cartesian refbit. In the finite copy scenario, we design machines that approximately break down Bell states of higher spins into Cartesian refbits, as well as machines that approximately implement the inverse process. In addition, we establish a quantitative link between the conversion of Bell states and the simulation of unitary gates, showing that the fidelity of probabilistic state conversion provides upper and lower bounds on the fidelity of deterministic gate simulation. The result holds not only for rotation gates, but also to all sets of gates that form finite-dimensional representations of compact groups. For rotation gates, we show how rotations on a system of given spin can simulate rotations on a system of different spin.

quant-ph↗