arXiv subjects

Yang Yu

Publications and source records attributed to Yang Yu.

At least 451 records · Page 25Linked to original sources

arXiv2018

Spin-valley-controlled photonic topological insulator

The discovery of photonic topological insulators (PTIs) has opened the door to fundamentally new topological states of light.Current time-reversal-invariant PTIs emulate either the quantum spin Hall (QSH) effect or the quantum valley Hall (QVH) effect in condensed-matter systems, in order to achieve topological transport of photons whose propagation is predetermined by either 'photonic pseudospin' (abbreviated as 'spin') or valley. Here we demonstrate a new class of PTIs, whose topological phase is not determined solely by spin or valley, but is controlled by the competition between their induced gauge fields. Such a competition is enabled by tuning the strengths of spin-orbit coupling (SOC) and inversion-symmetry breaking in a single PTI. An unprecedented topological transition between QSH and QVH phases that is hard to achieve in condensed-matter systems is demonstrated. Our study merges the emerging fields of spintronics and valleytronics in the same photonic platform, and offers novel PTIs with reconfigurable topological phases.

physics.optics↗

arXiv2018

Multi-Layered Gradient Boosting Decision Trees

Multi-layered representation is believed to be the key ingredient of deep neural networks especially in cognitive tasks like computer vision. While non-differentiable models such as gradient boosting decision trees (GBDTs) are the dominant methods for modeling discrete or tabular data, they are hard to incorporate with such representation learning ability. In this work, we propose the multi-layered GBDT forest (mGBDTs), with an explicit emphasis on exploring the ability to learn hierarchical representations by stacking several layers of regression GBDTs as its building block. The model can be jointly trained by a variant of target propagation across layers, without the need to derive back-propagation nor differentiability. Experiments and visualizations confirmed the effectiveness of the model in terms of performance and representation learning ability.

cs.LG↗

arXiv2018

Virtual-Taobao: Virtualizing Real-world Online Retail Environment for Reinforcement Learning

Applying reinforcement learning in physical-world tasks is extremely challenging. It is commonly infeasible to sample a large number of trials, as required by current reinforcement learning methods, in a physical environment. This paper reports our project on using reinforcement learning for better commodity search in Taobao, one of the largest online retail platforms and meanwhile a physical environment with a high sampling cost. Instead of training reinforcement learning in Taobao directly, we present our approach: first we build Virtual Taobao, a simulator learned from historical customer behavior data through the proposed GAN-SD (GAN for Simulating Distributions) and MAIL (multi-agent adversarial imitation learning), and then we train policies in Virtual Taobao with no physical costs in which ANC (Action Norm Constraint) strategy is proposed to reduce over-fitting. In experiments, Virtual Taobao is trained from hundreds of millions of customers' records, and its properties are compared with the real environment. The results disclose that Virtual Taobao faithfully recovers important properties of the real environment. We also show that the policies trained in Virtual Taobao can have significantly superior online performance to the traditional supervised approaches. We hope our work could shed some light on reinforcement learning applications in complex physical environments.

cs.AI↗

arXiv2018

Reinforcement Learning to Rank in E-Commerce Search Engine: Formalization, Analysis, and Application

In e-commerce platforms such as Amazon and TaoBao, ranking items in a search session is a typical multi-step decision-making problem. Learning to rank (LTR) methods have been widely applied to ranking problems. However, such methods often consider different ranking steps in a session to be independent, which conversely may be highly correlated to each other. For better utilizing the correlation between different ranking steps, in this paper, we propose to use reinforcement learning (RL) to learn an optimal ranking policy which maximizes the expected accumulative rewards in a search session. Firstly, we formally define the concept of search session Markov decision process (SSMDP) to formulate the multi-step ranking problem. Secondly, we analyze the property of SSMDP and theoretically prove the necessity of maximizing accumulative rewards. Lastly, we propose a novel policy gradient algorithm for learning an optimal ranking policy, which is able to deal with the problem of high reward variance and unbalanced reward distribution of an SSMDP. Experiments are conducted in simulation and TaoBao search engine. The results demonstrate that our algorithm performs much better than online LTR methods, with more than 40% and 30% growth of total transaction amount in the simulation and the real application, respectively.

cs.LG↗

arXiv2018

Electrically Induced Photonic and Acoustic Quantum Effect From Liquid Metal Droplets in Aqueous Solution

So far, several macroscopic quantum phenomena have been discovered in the Josephson junction. Through introducing such a structure with a liquid membrane sandwiched between two liquid metal electrodes, we had ever observed a lighting and sound phenomenon which was explained before as discharge plasma. In fact, such an effect also belongs to a quantum process. It is based on this conceiving, we proposed here that an electrically controllable method can thus be established to generate and manipulate as much photonic quantum as desired. We attributed such electrically induced lighting among liquid metal droplets immersed inside aqueous solution as photonic quantum effect. Our experiments clarified that a small electrical voltage would be strong enough to trigger blue-violet light and sound inside the aqueous solution system. Meanwhile, thermal heat is released, and chemical reaction occurs over the solution. From an alternative way which differs from former effort in interpreting such effect as discharge plasma, we treated this process as a quantum one and derived new conceptual equations to theoretically quantify this phenomenon in light of quantum mechanics principle. It can be anticipated that given specific designing, such spontaneously generated tremendous quantum can be manipulated to entangle together which would possibly help mold functional elements for developing future quantum computing or communication system. With superior adaptability than that of the conventional rigid junction, the present electro-photonic quantum generation system made of liquid metal droplets structure could work in solution, room temperature situation and is easy to be adjusted. It suggests a macroscopic way to innovate the classical strategies and technologies in generating quantum as frequently adopted in classical quantum engineering area.

physics.app-ph↗

arXiv2018

Knowledge Base Relation Detection via Multi-View Matching

Relation detection is a core component for Knowledge Base Question Answering (KBQA). In this paper, we propose a KB relation detection model via multi-view matching which utilizes more useful information extracted from question and KB. The matching inside each view is through multiple perspectives to compare two input texts thoroughly. All these components are designed in an end-to-end trainable neural network model. Experiments on SimpleQuestions and WebQSP yield state-of-the-art results.

cs.AI↗

arXiv2018

Topological Maxwell Metal Bands in a Superconducting Qutrit

We experimentally explore the topological Maxwell metal bands by mapping the momentum space of condensed-matter models to the tunable parameter space of superconducting quantum circuits. An exotic band structure that is effectively described by the spin-1 Maxwell equations is imaged. Three-fold degenerate points dubbed Maxwell points are observed in the Maxwell metal bands. Moreover, we engineer and observe the topological phase transition from the topological Maxwell metal to a trivial insulator, and report the first experiment to measure the Chern numbers that are higher than one.

quant-ph↗

arXiv2018

WikiRank: Improving Keyphrase Extraction Based on Background Knowledge

Keyphrase is an efficient representation of the main idea of documents. While background knowledge can provide valuable information about documents, they are rarely incorporated in keyphrase extraction methods. In this paper, we propose WikiRank, an unsupervised method for keyphrase extraction based on the background knowledge from Wikipedia. Firstly, we construct a semantic graph for the document. Then we transform the keyphrase extraction problem into an optimization problem on the graph. Finally, we get the optimal keyphrase set to be the output. Our method obtains improvements over other state-of-art models by more than 2% in F1-score.

cs.CL↗

arXiv2018

Accelerating E-Commerce Search Engine Ranking by Contextual Factor Selection

In industrial large-scale search systems, such as Taobao.com search for commodities, the quality of the ranking result is getting continually improved by introducing more factors from complex procedures, e.g., deep neural networks for extracting image factors. Meanwhile, the increasing of the factors demands more computation resource and raises the system response latency. It has been observed that a search instance usually requires only a small set of effective factors, instead of all factors. Therefore, removing ineffective factors significantly improves the system efficiency. This paper studies the \emph{Contextual Factor Selection} (CFS), which selects only a subset of effective factors for every search instance, for a well balance between the search quality and the response latency. We inject CFS into the search engine ranking score to accelerate the engine, considering both ranking effectiveness and efficiency. The learning of the CFS model involves a combinatorial optimization, which is transformed as a sequential decision-making problem. Solving the problem by reinforcement learning, we propose the RankCFS, which has been assessed in an off-line environment as well as a real-world on-line environment (Taobao.com). The empirical results show that, the proposed CFS approach outperforms several existing supervised/unsupervised methods for feature selection in the off-line environment, and also achieves significant real-world performance improvement, in term of service latency, in daily test as well as Singles' Day Shopping Festival in $2017$.

cs.IR↗

arXiv2018

Emulating topological chiral magnetic effects in artificial Weyl semimetals

We realized highly tunable Weyl semimetal-bands and subsequently emulated the topological chiral magnetic effects in superconducting quantum circuits. Driving the superconducting quantum circuits with elaborately designed microwave fields, we mapped the momentum space of a lattice to the parameter space, realizing the Hamiltonian of a Weyl semimetal. By measuring the energy spectrum, we directly imaged the Weyl points of cubic lattices, whose topological winding numbers were further determined from the Berry curvature measurement. In particular, we used an additional microwave field to produce a momentum-dependent chemical potential, from which the chiral magnetic topological current was extracted in the presence of an artificial magnetic field. This pure topological current is proportional to the magnetic field, which is in contrast to the famous Ampere's law, and may have significant impacts on topological materials and quantum devices.

cond-mat.mes-hall↗

arXiv2018

Tunneling Neural Perception and Logic Reasoning through Abductive Learning

Perception and reasoning are basic human abilities that are seamlessly connected as part of human intelligence. However, in current machine learning systems, the perception and reasoning modules are incompatible. Tasks requiring joint perception and reasoning ability are difficult to accomplish autonomously and still demand human intervention. Inspired by the way language experts decoded Mayan scripts by joining two abilities in an abductive manner, this paper proposes the abductive learning framework. The framework learns perception and reasoning simultaneously with the help of a trial-and-error abductive process. We present the Neural-Logical Machine as an implementation of this novel learning framework. We demonstrate that--using human-like abductive learning--the machine learns from a small set of simple hand-written equations and then generalizes well to complex equations, a feat that is beyond the capability of state-of-the-art neural network models. The abductive learning framework explores a new direction for approaching human-level learning ability.

cs.AI↗

arXiv2018

Demonstration of Hopf-link semimetal bands with superconducting circuits

Hopf-link semimetals exhibit exotic gapless band structures with fascinating topological properties, which have never been observed in nature. Here we demonstrate nodal lines with topological form of Hopf-link chains in artificial semimetal-bands. Driving superconducting quantum circuits with elaborately designed microwave fields, we mapped the momentum space of a lattice to a parameter space of the Hamiltonian for a Hopf-link semimetal. By measuring the energy spectrum, we directly imaged nodal lines in cubic lattices. By tuning the driving fields, we adjusted various parameters of Hamiltonian. Important topological features, such as link-unlink topological transitions and the robustness of the Hopf-link chain structure were investigated. Moreover, we extracted the linking number by detecting the Berry phase associated with different loops encircling nodal lines. This topological invariant clearly reveals the nontrivial topology of the Hopf-link semimetal. Our results provide knowledge for developing new materials and quantum devices.

quant-ph↗

arXiv2017

Sorting and Routing Topologically Protected Edge States by their Valleys

We experimentally demonstrate and characterize a photonic platform containing heterogeneous topological phases. This platform is shown to support valley-spin locked edge states that can be routed based on their spin and valley degrees of freedom. The routing is accomplished by changing the topology of the valley photonic crystal embedded inside a spin-Hall photonic crystal.

physics.app-ph↗

arXiv2017

Two-photon driven Kerr resonator for quantum annealing with three-dimensional circuit QED

We propose a realizable circuit QED architecture for engineering states of a superconducting resonator off-resonantly coupled to an ancillary superconducting qubit. The qubit-resonator dispersive interaction together with a microwave drive applied to the qubit gives rise to a Kerr resonator with two-photon driving that enables us to efficiently engineer the quantum state of the resonator such as generation of the Schrodinger cat states for resonator-based universal quantum computation. Moreover, the presented architecture is easily scalable for solving optimization problem mapped into the Ising spin glass model, and thus served as a platform for quantum annealing. Although various scalable architecture with superconducting qubits have been proposed for realizing quantum annealer, the existing annealers are currently limited to the coherent time of the qubits. Here, based on the protocol for realizing two-photon driven Kerr resonator in three-dimensional circuit QED (3D cQED), we propose a flexible and scalable hardware for implementing quantum annealer that combines the advantage of the long coherence times attainable in 3D cQED and the recently proposed resonator based Lechner-Hauke-Zoller (LHZ) scheme. In the proposed resonator based LHZ annealer, each spin is encoded in the subspace formed by two coherent state of 3D microwave superconducting resonator with opposite phase, and thus the fully-connected Ising model is mapped onto the network of the resonator with local tunable three-resonator interaction. This hardware architecture provides a promising physical platform for realizing quantum annealer with improved coherence.

quant-ph↗

arXiv2017

Quantum simulation of general semi-classical Rabi model beyond strong driving regime

We propose a scheme to simulate the interaction between a two-level system and a classical light field. Under the transversal driving of two microwave tones, the system Hamiltonian is identical to that of the general semi-classical Rabi model. We experimentally realize this Hamiltonian with a superconducting transmon qubit. By tuning the strength, phase and frequency of the two microwave driving fields, we simulate the quantum dynamics from weak to extremely strong driving regime. The resulting evolutions gradually deviate from the normal sinusoidal Rabi oscillations with increasing driving strength, in accordance with the predictions of the general semi-classical Rabi model far beyond the weak driving limit. Our scheme provides an effective approach to investigate the extremely strong interaction between a two-level system and a classical light field. Such strong interactions are usually inaccessible in experiments.

quant-ph↗

arXiv2017

Constraints on the perturbed mutual motion in Didymos due to impact-induced deformation of its primary after the DART impact

Binary near-Earth asteroid (65803) Didymos is the target of the proposed NASA Double Asteroid Redirection Test (DART), part of the Asteroid Impact & Deflection Assessment (AIDA) mission concept. In this mission, the DART spacecraft is planned to impact the secondary body of Didymos, perturbing mutual dynamics of the system. The primary body is currently rotating at a spin period close to the spin barrier of asteroids, and materials ejected from the secondary due to the DART impact are likely to reach the primary. These conditions may cause the primary to reshape, due to landslides, or internal deformation, changing the permanent gravity field. Here, we propose that if shape deformation of the primary occurs, the mutual orbit of the system would be perturbed due to a change in the gravity field. We use a numerical simulation technique based on the full two-body problem to investigate the shape effect on the mutual dynamics in Didymos after the DART impact. The results show that under constant volume, shape deformation induces strong perturbation in the mutual motion. We find that the deformation process always causes the orbital period of the system to become shorter. If surface layers with a thickness greater than ~0.4 m on the poles of the primary move down to the equatorial region due to the DART impact, a change in the orbital period of the system and in the spin period of the primary will be detected by ground-based measurement.

astro-ph.EP↗

arXiv2017

Structural analysis of rubble-pile asteroids applied to collisional evolution

Solar system small bodies come in a wide variety of shapes and sizes, which are achieved following very individual evolutional paths through billions of years. This paper focuses on the reshaping process of rubble-pile asteroids driven by meteorite impacts. In our study, numerous possible equilibrium configurations are obtained via Monte Carlo simulation, and the structural stability of these configurations is determined via eigen analysis of the geometric constructions. The eigen decomposition reveals a connection between the cluster's reactions and the types of external disturbance. Numerical simulations are performed to verify the analytical results. The gravitational N-body code pkdgrav is used to mimic the responses of the cluster under intermittent non-dispersive impacts. We statistically confirm that the stability index, the total gravitational potential and the volume of inertia ellipsoid show consistent tendency of variation. A common regime is found in which the clusters tend towards crystallization under intermittent impacts, i.e., only the configurations with high structural stability survive under the external disturbances. The results suggest the trivial non-disruptive impacts might play an important role in the rearrangement of the constituent blocks, which may strengthen these rubble piles and help to build a robust structure under impacts of similar magnitude. The final part of this study consists of systematic simulations over two parameters, the projectile momentum and the rotational speed of the cluster. The results show a critical value exists for the projectile momentum, as predicted by theory, below which all clusters become responseless to external disturbances; and the rotation proves to be significant for it exhibits an "enhancing" effect on loose-packed clusters, which coincides with the observation that several fast-spinning asteroids have low bulk densities.

astro-ph.EP↗

arXiv2017

Ejecta cloud from the AIDA space project kinetic impact on the secondary of a binary asteroid: II. fates and evolutionary dependencies

This paper presents a quantitative study of the evolution of the ejecta cloud released from a hypervelocity impact on a binary asteroid. We performed numerical simulations of the post-impact dynamics of the ejecta cloud in the framework of the current mission scenario of AIDA mission project. A grid search of launching sites of ejecta was defined over the globe of Didymoon, and considering a wide range of possible ejection speeds, we determined the dependency of ejecta fate on launching sites (projectile impact sites) and speeds. This range allows us to track all the complex cases that include different types of dynamical fates. Two major mechanisms are found to be working broadly during the post-ejection evolution of the ejecta cloud: 1) ejecta on mean motion resonance orbits with Didymoon produce long-term quasi-periodic showers onto Didymoon over at least a couple of weeks after the projectile impact, 2) ejecta on non-resonant orbits produce a rapid and high re-accretion flux. This rapid and high flux occurs just once because ejecta on such orbits leave the system unless they experience a collision during their first encounter. For the second part of this study, we performed full-scale simulations of the ejecta cloud released from 6 hypothetical impact sites. We considered two kinds of material composing Didymoon's subsurface and then combined a power-law size distribution of the ejecta with an ejection speed distribution. We find that the ejecta cloud evolution can be divided in two periods. It starts with a first violent period (<10 hr) with fast re-accretion or ejection from the system. A second period is found to be more sensitive to the launching site than the first one. During this second period, ejecta will either re-accrete or being ejected from the system, depending both on their sizes and on their average survival time in close proximity of the binary components.

astro-ph.EP↗