arXiv subjects

Li Zhao

Publications and source records attributed to Li Zhao.

At least 181 records · Page 10Linked to original sources

arXiv2020

Suphx: Mastering Mahjong with Deep Reinforcement Learning

Artificial Intelligence (AI) has achieved great success in many domains, and game AI is widely regarded as its beachhead since the dawn of AI. In recent years, studies on game AI have gradually evolved from relatively simple environments (e.g., perfect-information games such as Go, chess, shogi or two-player imperfect-information games such as heads-up Texas hold'em) to more complex ones (e.g., multi-player imperfect-information games such as multi-player Texas hold'em and StartCraft II). Mahjong is a popular multi-player imperfect-information game worldwide but very challenging for AI research due to its complex playing/scoring rules and rich hidden information. We design an AI for Mahjong, named Suphx, based on deep reinforcement learning with some newly introduced techniques including global reward prediction, oracle guiding, and run-time policy adaptation. Suphx has demonstrated stronger performance than most top human players in terms of stable rank and is rated above 99.99% of all the officially ranked human players in the Tenhou platform. This is the first time that a computer program outperforms most top human players in Mahjong.

cs.AI↗

arXiv2020

A Vision of C-V2X: Technologies, Field Testing and Challenges with Chinese Development

C-V2X (Cellular Vehicle-to-Everything) is the important enabling technology for autonomous driving and intelligent transportation systems. It evolves from LTE (Long Term Evolution)-V2X to NR (New Radio)-V2X, which will coexist and be complementary with each other to provide low latency, high reliability, and high throughput communications for various C-V2X applications. In this article, a vision of C-V2X is presented. The requirements of the basic road safety and advanced applications, the architecture, the key technologies, and the standards of C-V2X are introduced, highlighting the technical evolution path from LTE-V2X to NR-V2X. Especially, based on the continual and active promotion of C-V2X research, field testing and development in China, the related works and progresses are also presented. Lastly, the trends of C-V2X applications with technical challenges are envisioned.

eess.SP↗

arXiv2020

Trimming the Sail: A Second-order Learning Paradigm for Stock Prediction

Nowadays, machine learning methods have been widely used in stock prediction. Traditional approaches assume an identical data distribution, under which a learned model on the training data is fixed and applied directly in the test data. Although such assumption has made traditional machine learning techniques succeed in many real-world tasks, the highly dynamic nature of the stock market invalidates the strict assumption in stock prediction. To address this challenge, we propose the second-order identical distribution assumption, where the data distribution is assumed to be fluctuating over time with certain patterns. Based on such assumption, we develop a second-order learning paradigm with multi-scale patterns. Extensive experiments on real-world Chinese stock data demonstrate the effectiveness of our second-order learning paradigm in stock prediction.

q-fin.ST↗

arXiv2020

Real-time nanodiamond thermometry probing in-vivo thermogenic responses

Real-time temperature monitoring inside living organisms provides a direct measure of their biological activities, such as homeostatic thermoregulation and energy metabolism. However, it is challenging to reduce the size of bio-compatible thermometers down to submicrometers despite their potential applications for the thermal imaging of subtissue structures with single-cell resolution. Light-emitting nanothermometers that remotely sense temperature via optical signals exhibit considerable potential in such \textit{in-vivo} high-spatial-resolution thermometry. Here, using quantum nanothermometers based on optically accessible electron spins in nanodiamonds (NDs), we demonstrate \textit{in-vivo} real-time temperature monitoring inside \textit{Caenorhabditis elegans} (\textit{C. elegans}) worms. We developed a thermometry system that can measure the temperatures of movable NDs inside live adult worms with a precision of $\pm 0.22^{\circ}{\rm C}$. Using this system, we determined the increase in temperature based on the thermogenic responses of the worms during the chemical stimuli of mitochondrial uncouplers. Our technique demonstrates sub-micrometer localization of real-time temperature information in living animals and direct identification of their pharmacological thermogenesis. The results obtained facilitate the development of a method to probe subcellular temperature variation inside living organisms and may allow for quantification of their biological activities based on their energy expenditures.

q-bio.QM↗

arXiv2019

$q$-form field on a $p$-brane with codimension two

In this paper, by performing a general Kaluza-Klein (KK) decomposition, we obtain a gauge invariant effective action for a bulk massless $q$-form field on a $p$-brane with codimension two. There appear four types of KK modes: two $(q-1)$-forms and one $(q-2)$-form in addtion to the ordinary $q$-form, which are essential for the gauge invariance. Due to the two extra dimensions, we find eight Schrödinger-like equations for the four modes and their mass spectra are closely related. Moreover, via this decomposition mechanism, the Hodge duality in the bulk naturally induces four coupled dualities on the brane, which guarantees that the physical equavalence of bulk dual fields is preserved under the dimensional reduction.

hep-th↗

arXiv2019

Design and commissioning of a 600 L Time Projection Chamber with Microbulk Micromegas

We report the design, construction, and initial commissioning results of a large high pressure gaseous Time Projection Chamber (TPC) with Micromegas modules for charge readout. The detector vessel has an inner volume of about 600 L and an active volume of 270 L. At 10 bar operating pressure, the active volume contains about 20 kg of xenon gas and can image charged particle tracks. Drift electrons are collected by the charge readout plane, which accommodates a tessellation of seven Micromegas modules. Each of the Micromegas covers a square of 20 cm by 20 cm. A new type of Microbulk Micromegas is chosen for this application due to its good gain uniformity and low radioactive contamination. Initial commissioning results with 1 Micromegas module running with 1 bar argon and isobutane gas mixture and 5 bar xenon and trimethylamine (TMA) gas mixture are reported. We also recorded extended background tracks from cosmic ray events and highlighted the unique tracking feature of this gaseous TPC.

physics.ins-det↗

arXiv2019

Kalb-Ramond field localization on a de Sitter Thick Brane

In this paper, we study the localization of Kalb-Ramond (KR) tensorial gauge field on a non-flat de Sitter thick brane. The localization and resonance of KR gauge field are discussed for three kinds of couplings. For the first coupling there is no localized tensorial zero mode. For the other two couplings, the zero mode of KR field can be localized under certain condition. There are resonant KK modes on the thick brane for the third case. Furthermore, we mainly analyze the effects of three parameters on the localization and resonant mode for KR field.

gr-qc↗

arXiv2019

Distributional Reward Decomposition for Reinforcement Learning

Many reinforcement learning (RL) tasks have specific properties that can be leveraged to modify existing RL algorithms to adapt to those tasks and further improve performance, and a general class of such properties is the multiple reward channel. In those environments the full reward can be decomposed into sub-rewards obtained from different channels. Existing work on reward decomposition either requires prior knowledge of the environment to decompose the full reward, or decomposes reward without prior knowledge but with degraded performance. In this paper, we propose Distributional Reward Decomposition for Reinforcement Learning (DRDRL), a novel reward decomposition algorithm which captures the multiple reward channel structure under distributional setting. Empirically, our method captures the multi-channel structure and discovers meaningful reward decomposition, without any requirements on prior knowledge. Consequently, our agent achieves better performance than existing methods on environments with multiple reward channels.

cs.LG↗

arXiv2019

Thick brane in reduced Horndeski theory

Horndeski theory is the most general scalar-tensor theory retaining second-order field equations, although the action includes higher-order terms. This is achieved by a special choice of coupling constants. In this paper, we investigate thick brane system in reduced Horndeski theory, especially the effect of the non-minimal derivative coupling on thick brane. First, the equations of motion are presented and a set of analytic background solutions are obtained. Then, to investigate the stability of the background scalar profile, we present a novel canonically normalized method, and show that although the original background scalar field is unstable, the canonical one is stable. The stability of the thick brane under tensor perturbation is also considered. It is shown that the tachyon is absent and the graviton zero mode can be localized on the brane. The localized graviton zero mode recovers the four-dimensional Newtonian potential and the presence of the non-minimal derivative coupling results in a splitting of its wave function. The correction of the massive graviton KK modes to the Newtonian potential is also analyzed briefly.

gr-qc↗

arXiv2019

An Improved Evaluation of the Neutron Background in the PandaX-II Experiment

In dark matter direct detection experiments, neutron is a serious source of background, which can mimic the dark matter-nucleus scattering signals. In this paper, we present an improved evaluation of the neutron background in the PandaX-II dark matter experiment by a novel approach. Instead of fully relying on the Monte Carlo simulation, the overall neutron background is determined from the neutron-induced high energy signals in the data. In addition, the probability of producing a dark-matter-like background per neutron is evaluated with a complete Monte Carlo generator, where the correlated emission of neutron(s) and $γ$(s) in the ($α$, n) reactions and spontaneous fissions is taken into consideration. With this method, the neutron backgrounds in the Run 9 (26-ton-day) and Run 10 (28-ton-day) data sets of PandaX-II are estimated to be 0.66$\pm$0.24 and 0.47$\pm$0.25 events, respectively.

hep-ex↗

arXiv2019

Searching for Neutrino-less Double Beta Decay of $^{136}$Xe with PandaX-II Liquid Xenon Detector

We report the Neutrino-less Double Beta Decay (NLDBD) search results from PandaX-II dual-phase liquid xenon time projection chamber. The total live time used in this analysis is 403.1 days from June 2016 to August 2018. With NLDBD-optimized event selection criteria, we obtain a fiducial mass of 219 kg of natural xenon. The accumulated xenon exposure is 242 kg$\cdot$yr, or equivalently 22.2 kg$\cdot$yr of $^{136}$Xe exposure. At the region around $^{136}$Xe decay Q-value of 2458 keV, the energy resolution of PandaX-II is 4.2%. We find no evidence of NLDBD in PandaX-II and establish a lower limit for decay half-life of 2.4 $ \times 10^{23} $ yr at the 90% confidence level, which corresponds to an effective Majorana neutrino mass $m_{ββ} < (1.3 - 3.5)$ eV. This is the first NLDBD result reported from a dual-phase xenon experiment.

hep-ex↗

arXiv2019

Thomson backscattering in combined two laser and magnetic field

The Thomson backscattering of an electron moving in combined fields is studied by a dynamically assisted mechanism. The combined fields are composed of two co-propagating laser fields and a magnetic field, where the first laser field is strong and low-frequency while the second is weak and high-frequency, relatively. The dependence of fundamental frequency of emission on the ratio of incident laser high-to-low frequency is presented and the spectrum of backscattering is obtained. It is found that, with a magnetic field, the peak of the spectrum and the corresponding radiation frequency are significantly larger in case of two-laser than that in case of only one laser. They are also improved obviously as the frequency of the weak laser field. Another finding is the nonlinear correlation between the emission intensity of the backscattering and the intensity of the weak laser field. These results provide a new possibility to adjust and control the spectrum by changing the ratios of frequency and intensity of the two laser fields.

physics.optics↗

arXiv2019

PandaX-II Constraints on Spin-Dependent WIMP-Nucleon Effective Interactions

We present PandaX-II constraints on candidate WIMP-nucleon effective interactions involving the nucleon or WIMP spin, including, in addition to standard axial spin-dependent (SD) scattering, various couplings among vector and axial currents, magnetic and electric dipole moments, and tensor interactions. The data set corresponding to a total exposure of 54-ton-days is reanalyzed to determine constraints as a function of the WIMP mass and isospin coupling. We obtain WIMP-nucleon cross section bounds of $\rm 1.6 \times 10^{-41} cm^2$ and $\rm 9.0 \times 10^{-42} cm^2$ ($90\%$ c.l.) for neutron-only SD and tensor coupling, respectively, for a mass $M_\mathrm{WIMP} \sim {\rm 40~GeV}/c^2$. The SD limits are the best currently available for $M_\mathrm{WIMP} > {\rm 40~GeV}/c^2$. We show that PandaX-II has reached a sensitivity sufficient to probe a variety of other candidate spin-dependent interactions at the weak scale.

hep-ex↗

arXiv2018

Linear Stability of $f(R,ϕ,X)$ Thick Branes: Tensor Perturbations

We explore thick branes in $f(R,ϕ,X)$ gravity. We obtain the linear tensor perturbation equation of $f(R,ϕ,X)$ branes and show that the branes are stable against the tensor perturbations under the condition of $\frac{\partial f(R,ϕ,X)}{\partial R}>0$. In order to obtain thick brane solutions of the fourth-order field equations in this theory, we employ the reconstruction technique. We get exact solutions of the specific $f(R,ϕ,X)$ thick brane generated by a non-canonical scalar field. It is shown that the zero mode of the graviton for the thick brane is localized under certain conditions. This implies that the four-dimensional Newtonian potential is recovered on the brane. The effects of the Kaluza-Klein modes of the graviton for the $f(R,ϕ,X)$ thick brane are also discussed.

hep-th↗

arXiv2018

Angular Distributions of Thomson Scattering in Combined Laser and Magnetic fields

Angular distributions of Thomson scattering are researched in the combined fields with a circularly polarization of laser field and a strong uniform magnetic field. The trajectories of the electron and the dependence of it on the initial phase are also given. It is found that the angular distributions with respect to the azimuthal angle show twofold symmetry whatever the laser intensity, the order of harmonics, the resonance parameter, and the initial axial momentum are. On the other hand, the radiation with respect to the polar angle is mainly distributed in two regions which are roughly symmetric of the laser propagation. In addition, the larger the laser intensity, the resonance parameters, the initial axial momentum are, the closer radiation is to the laser propagation direction. Besides, a new possibility of X-ray production is indicated. That is to say, with appropriate choice of laser and electron parameters, the high frequency part of the Thomson scattering radiation can reach the frequency range of X-ray (10^17 Hz - 10^18 Hz).

physics.plasm-ph↗

arXiv2018

Adversarial Neural Machine Translation

In this paper, we study a new learning paradigm for Neural Machine Translation (NMT). Instead of maximizing the likelihood of the human translation as in previous works, we minimize the distinction between human translation and the translation given by an NMT model. To achieve this goal, inspired by the recent success of generative adversarial networks (GANs), we employ an adversarial training architecture and name it as Adversarial-NMT. In Adversarial-NMT, the training of the NMT model is assisted by an adversary, which is an elaborately designed Convolutional Neural Network (CNN). The goal of the adversary is to differentiate the translation result generated by the NMT model from that by human. The goal of the NMT model is to produce high quality translations so as to cheat the adversary. A policy gradient method is leveraged to co-train the NMT model and the adversary. Experimental results on English$\rightarrow$French and German$\rightarrow$English translation tasks show that Adversarial-NMT can achieve significantly better translation quality than several strong baselines.

cs.CL↗

arXiv2018

Error Reduction Network for DBLSTM-based Voice Conversion

So far, many of the deep learning approaches for voice conversion produce good quality speech by using a large amount of training data. This paper presents a Deep Bidirectional Long Short-Term Memory (DBLSTM) based voice conversion framework that can work with a limited amount of training data. We propose to implement a DBLSTM based average model that is trained with data from many speakers. Then, we propose to perform adaptation with a limited amount of target data. Last but not least, we propose an error reduction network that can improve the voice conversion quality even further. The proposed framework is motivated by three observations. Firstly, DBLSTM can achieve a remarkable voice conversion by considering the long-term dependencies of the speech utterance. Secondly, DBLSTM based average model can be easily adapted with a small amount of data, to achieve a speech that sounds closer to the target. Thirdly, an error reduction network can be trained with a small amount of training data, and can improve the conversion quality effectively. The experiments show that the proposed voice conversion framework is flexible to work with limited training data and outperforms the traditional frameworks in both objective and subjective evaluations.

eess.AS↗

arXiv2018

Reinforcement Learning for Relation Classification from Noisy Data

Existing relation classification methods that rely on distant supervision assume that a bag of sentences mentioning an entity pair are all describing a relation for the entity pair. Such methods, performing classification at the bag level, cannot identify the mapping between a relation and a sentence, and largely suffers from the noisy labeling problem. In this paper, we propose a novel model for relation classification at the sentence level from noisy data. The model has two modules: an instance selector and a relation classifier. The instance selector chooses high-quality sentences with reinforcement learning and feeds the selected sentences into the relation classifier, and the relation classifier makes sentence level prediction and provides rewards to the instance selector. The two modules are trained jointly to optimize the instance selection and relation classification processes. Experiment results show that our model can deal with the noise of data effectively and obtains better performance for relation classification at the sentence level.

cs.IR↗