arXiv subjects

Yu Zeng

Publications and source records attributed to Yu Zeng.

At least 91 records · Page 5Linked to original sources

arXiv2020

Delay Adaptation Method for Relay Assisted Optical Wireless Systems

In this paper, we investigate optical wireless repeaters as relay terminals between a transmitter and a user in an Infrared Optical Wireless Communication (IROWC) system. A delay adaptation method is introduced to solve the problem of irregular signal arrival time from different relay terminals. Three different relay terminal deployment scenarios were investigated in a typical two-phase relay IROWC system with the proposed delay adaptation method. The simulation results indicate that the proposed system has better impulse response compared to the conventional system and that the root-mean-square delay spread of the relay system with the delay adaptation method is on average 30% less than the conventional system.

eess.SP↗

arXiv2020

Diagonal entropy in many-body systems: Volume effect and quantum phase transitions

We investigate the diagonal entropy(DE) of the ground state for quantum many-body systems, including the XY model and the Ising model with next nearest neighbour interactions. We focus on the DE of a subsystem of L continuous spins. We show that the DE in many-body systems, regardless of integrability, can be represented as a volume term plus a logarithmic correction and a constant offset. Quantum phase transition points can be explicitly identified by the three coefficients thereof. Besides, by combining entanglement entropy and the relative entropy of quantum coherence, as two celebrated representatives of quantumness, we simply obtain the DE, which naturally has the potential to reveal the information of quantumness. More importantly, the DE is concerning only the diagonal form of the ground state reduced density matrix, making it feasible to measure in real experiments, and therefore it has immediate applications in demonstrating quantum supremacy on state-of-the-art quantum simulators.

quant-ph↗

arXiv2019

Statistics of Deep Generated Images

Here, we explore the low-level statistics of images generated by state-of-the-art deep generative models. First, Variational auto-encoder (VAE~\cite{kingma2013auto}), Wasserstein generative adversarial network (WGAN~\cite{arjovsky2017wasserstein}) and deep convolutional generative adversarial network (DCGAN~\cite{radford2015unsupervised}) are trained on the ImageNet dataset and a large set of cartoon frames from animations. Then, for images generated by these models as well as natural scenes and cartoons, statistics including mean power spectrum, the number of connected components in a given image area, distribution of random filter responses, and contrast distribution are computed. Our analyses on training images support current findings on scale invariance, non-Gaussianity, and Weibull contrast distribution of natural scenes. We find that although similar results hold over cartoon images, there is still a significant difference between statistics of natural scenes and images generated by VAE, DCGAN and WGAN models. In particular, generated images do not have scale invariant mean power spectrum magnitude, which indicates existence of extra structures in these images. Inspecting how well the statistics of deep generated images match the known statistical properties of natural images, such as scale invariance, non-Gaussianity, and Weibull contrast distribution, can a) reveal the degree to which deep learning models capture the essence of the natural scenes, b) provide a new dimension to evaluate models, and c) allow possible improvement of image generative models (e.g., via defining new loss functions).

cs.CV↗

arXiv2019

Joint Learning of Saliency Detection and Weakly Supervised Semantic Segmentation

Existing weakly supervised semantic segmentation (WSSS) methods usually utilize the results of pre-trained saliency detection (SD) models without explicitly modeling the connections between the two tasks, which is not the most efficient configuration. Here we propose a unified multi-task learning framework to jointly solve WSSS and SD using a single network, \ie saliency, and segmentation network (SSNet). SSNet consists of a segmentation network (SN) and a saliency aggregation module (SAM). For an input image, SN generates the segmentation result and, SAM predicts the saliency of each category and aggregating the segmentation masks of all categories into a saliency map. The proposed network is trained end-to-end with image-level category labels and class-agnostic pixel-level saliency labels. Experiments on PASCAL VOC 2012 segmentation dataset and four saliency benchmark datasets show the performance of our method compares favorably against state-of-the-art weakly supervised segmentation methods and fully supervised saliency detection methods.

cs.CV↗

arXiv2019

Observation of dynamical quantum phase transition by a superconducting qubit simulation

A dynamical quantum phase transition can occur during time evolution of sudden quenched quantum systems across a phase transition. It corresponds to the nonanalytic behavior at a critical time of the rate function of the quantum state return amplitude, analogous to nonanalyticity of the free energy density at the critical temperature in macroscopic systems. A variety of many-body systems can be represented in momentum space as a spin-1/2 state evolving on the Bloch sphere, where each momentum mode is decoupled and thus can be simulated independently by a single qubit. Here, we report the observation of a dynamical quantum phase transition in a superconducting qubit simulation of the quantum quench dynamics of many-body systems. We take the Ising model with a transverse field as an example for demonstration. In our experiment, the spin state, which is initially polarized longitudinally, evolves based on a Hamiltonian with adjustable parameters depending on the momentum and strength of the transverse magnetic field. The time evolving quantum state is read out by state tomography. Evidence of dynamical quantum phase transitions, such as paths of time evolution states on the Bloch sphere, non-analytic behavior of the dynamical free energy and the emergence of Skyrmion lattice in momentum-time space, is observed. The experimental data agrees well with theoretical and numerical calculations. The experiment demonstrates for the first time explicitly the topological invariant, both topologically trivial and non-trivial, for dynamical quantum phase transitions. Our results show that the quantum phase transitions of this class of many-body systems can be simulated successfully with a single qubit by varying certain control parameters over the corresponding momentum range.

quant-ph↗

arXiv2019

Energy Efficient virtualization framework for 5G F-RAN

Fog radio access network (F-RAN) and virtualisation are promising technologies for 5G networks. In F-RAN, the fog and cloud computing are integrated where the conventional C-RAN functions are diverged to the edge devices of radio access networks. F-RAN is adopted to mitigate the burden of front-haul and improve the end to end (E2E) latency. On other hand, virtualization and network function virtualization (NFV) are IT techniques that aim to convert the functions from hardware to software based functions. Many merits could be brought by the employment of NFV in mobile networks including a high degree of reliability, flexibility and energy efficiency. In this paper, a virtualization framework is introduced for F-RAN to improve the energy efficiency in 5G networks. In this framework, a gigabit passive optical network (GPON) is leveraged as a backbone network for the proposed F-RAN architecture where it connects several evolved nodes B (eNodeBs) via fibre cables. The energy-efficiency of the proposed F-RAN architecture has been investigated and compared with the conventional C-RAN architecture in two different scenarios using mixed integer linear programming (MILP) models. The MILP results indicate that on average a 30% power saving can be achieved by the F-RAN architecture compared with the C-RAN architecture.

cs.NI↗

arXiv2019

Multi-source weak supervision for saliency detection

The high cost of pixel-level annotations makes it appealing to train saliency detection models with weak supervision. However, a single weak supervision source usually does not contain enough information to train a well-performing model. To this end, we propose a unified framework to train saliency detection models with diverse weak supervision sources. In this paper, we use category labels, captions, and unlabelled data for training, yet other supervision sources can also be plugged into this flexible framework. We design a classification network (CNet) and a caption generation network (PNet), which learn to predict object categories and generate captions, respectively, meanwhile highlight the most important regions for corresponding tasks. An attention transfer loss is designed to transmit supervision signal between networks, such that the network designed to be trained with one supervision source can benefit from another. An attention coherence loss is defined on unlabelled data to encourage the networks to detect generally salient regions instead of task-specific regions. We use CNet and PNet to generate pixel-level pseudo labels to train a saliency prediction network (SNet). During the testing phases, we only need SNet to predict saliency maps. Experiments demonstrate the performance of our method compares favourably against unsupervised and weakly supervised methods and even some supervised methods.

cs.CV↗

arXiv2018

Automatic Seismic Salt Interpretation with Deep Convolutional Neural Networks

One of the most crucial tasks in seismic reflection imaging is to identify the salt bodies with high precision. Traditionally, this is accomplished by visually picking the salt/sediment boundaries, which requires a great amount of manual work and may introduce systematic bias. With recent progress of deep learning algorithm and growing computational power, a great deal of efforts have been made to replace human effort with machine power in salt body interpretation. Currently, the method of Convolutional neural networks (CNN) is revolutionizing the computer vision field and has been a hot topic in the image analysis. In this paper, the benefits of CNN-based classification are demonstrated by using a state-of-art network structure U-Net, along with the residual learning framework ResNet, to delineate salt body with high precision. Network adjustments, including the Exponential Linear Units (ELU) activation function, the Lovász-Softmax loss function, and stratified $K$-fold cross-validation, have been deployed to further improve the prediction accuracy. The preliminary result using SEG Advanced Modeling (SEAM) data shows good agreement between the predicted salt body and manually interpreted salt body, especially in areas with weak reflections. This indicates the great potential of applying CNN for salt-related interpretations.

physics.geo-ph↗

arXiv2018

Scalable Mutual Information Estimation using Dependence Graphs

The Mutual Information (MI) is an often used measure of dependency between two random variables utilized in information theory, statistics and machine learning. Recently several MI estimators have been proposed that can achieve parametric MSE convergence rate. However, most of the previously proposed estimators have the high computational complexity of at least $O(N^2)$. We propose a unified method for empirical non-parametric estimation of general MI function between random vectors in $\mathbb{R}^d$ based on $N$ i.i.d. samples. The reduced complexity MI estimator, called the ensemble dependency graph estimator (EDGE), combines randomized locality sensitive hashing (LSH), dependency graphs, and ensemble bias-reduction methods. We prove that EDGE achieves optimal computational complexity $O(N)$, and can achieve the optimal parametric MSE rate of $O(1/N)$ if the density is $d$ times differentiable. To the best of our knowledge EDGE is the first non-parametric MI estimator that can achieve parametric MSE rates with linear time complexity. We illustrate the utility of EDGE for the analysis of the information plane (IP) in deep learning. Using EDGE we shed light on a controversy on whether or not the compression property of information bottleneck (IB) in fact holds for ReLu and other rectification functions in deep neural networks (DNN).

cs.IT↗

arXiv2018

A Universal Quantum Computing Virtual Machine

A medium-scale quantum computer with full universal quantum computing capability is necessary for various practical aims and testing applications. Here we report a 34-qubit quantum virtual machine (QtVM) based on a medium server. Our QtVM can run quantum assembly language with graphic interfaces. The QtVM is implemented with single qubit rotation gate, single to multiple controlled NOT gates to realize the universal quantum computation. Remarkably, it can realize a series of basic functions, such as, the "if" conditional programming language based on single-shot projective measurement results, "for" iteration programming language, build in arithmetic calculation. The measurement can be single-shot and arbitrary number of multi-shot types. In addition, there is in principle no limitation on number of logic gates implemented for quantum computation. By using QtVM, we demonstrate the simulation of dynamical quantum phase transition of transverse field Ising model by quantum circuits, where 34 qubits with one million gates are realized. We also show the realization of programmable Shor algorithm for factoring 15 and 35.

quant-ph↗

arXiv2018

Application of Machine Learning in Rock Facies Classification with Physics-Motivated Feature Augmentation

With recent progress in algorithms and the availability of massive amounts of computation power, application of machine learning techniques is becoming a hot topic in the oil and gas industry. One of the most promising aspects to apply machine learning to the upstream field is the rock facies classification in reservoir characterization, which is crucial in determining the net pay thickness of reservoirs, thus a definitive factor in drilling decision making process. For complex machine learning tasks like facies classification, feature engineering is often critical. This paper shows the inclusion of physics-motivated feature interaction in feature augmentation can further improve the capability of machine learning in rock facies classification. We demonstrate this approach with the SEG 2016 machine learning contest dataset and the top winning algorithms. The improvement is roboust and can be $\sim5\%$ better than current existing best F-1 score, where F-1 is an evaluation metric used to quantify average prediction accuracy.

stat.ML↗

arXiv2018

Characterization of topological states via dual multipartite entanglement

We demonstrate that multipartite entanglement is able to characterize one-dimensional symmetry-protected topological order, which is witnessed by the scaling behavior of the quantum Fisher information of the ground state with respect to the spin operators defined in the dual lattice. We investigate an extended Kitaev chain with a $\mathbf{Z}$ symmetry identified equivalently by winding numbers and paired Majorana zero modes at each end. The topological phases with high winding numbers are detected by the scaling coefficient of the quantum Fisher information density with respect to generators in different dual lattices. Containing richer properties and more complex structures than bipartite entanglement, the dual multipartite entanglement of the topological state has promising applications in robust quantum computation and quantum metrology, and can be generalized to identify topological order in the Kitaev honeycomb model.

quant-ph↗

arXiv2017

Interfaces of high efficient kesterite Cu2ZnSnS(e)4 thin film solar cells

Cu2ZnSnS(e)4 (CZTS(e)) solar cells have attracted much attention due to the elemental abundance and the non-toxicity. However, the record efficiency of 12.6% for Cu2ZnSn(S,Se)4 (CZTSSe) solar cells is much lower than that of Cu(In,Ga)Se2 (CIGS) solar cells. One crucial reason is the recombination at interfaces. In recent years, large amount investigations have been done to analyze the interfacial problems and improve the interfacial properties via a variety of methods. This paper gives a review of progresses on interfaces of CZTS(e) solar cells, including: (1) the band alignment optimization at buffer/CZTS(e) interface, (2) tailoring the thickness of MoS(e)2 interfacial layers between CZTS(e) absorber and Mo back contact, (3) the passivation of rear interface, (4) the passivation of front interface, and (5) the etching of secondary phases.

physics.app-ph↗

arXiv2017

Emulating many-body localization with a superconducting quantum processor

The law of statistical physics dictates that generic closed quantum many-body systems initialized in nonequilibrium will thermalize under their own dynamics. However, the emergence of many-body localization (MBL) owing to the interplay between interaction and disorder, which is in stark contrast to Anderson localization that only addresses noninteracting particles in the presence of disorder, greatly challenges this concept because it prevents the systems from evolving to the ergodic thermalized state. One critical evidence of MBL is the long-time logarithmic growth of entanglement entropy, and a direct observation of it is still elusive due to the experimental challenges in multiqubit single-shot measurement and quantum state tomography. Here we present an experiment of fully emulating the MBL dynamics with a 10-qubit superconducting quantum processor, which represents a spin-1/2 XY model featuring programmable disorder and long-range spin-spin interactions. We provide essential signatures of MBL, such as the imbalance due to the initial nonequilibrium, the violation of eigenstate thermalization hypothesis, and, more importantly, the direct evidence of the long-time logarithmic growth of entanglement entropy. Our results lay solid foundations for precisely simulating the intriguing physics of quantum many-body systems on the platform of large-scale multiqubit superconducting quantum processors.

quant-ph↗

arXiv2017

An Unsupervised Game-Theoretic Approach to Saliency Detection

We propose a novel unsupervised game-theoretic salient object detection algorithm that does not require labeled training data. First, saliency detection problem is formulated as a non-cooperative game, hereinafter referred to as Saliency Game, in which image regions are players who choose to be "background" or "foreground" as their pure strategies. A payoff function is constructed by exploiting multiple cues and combining complementary features. Saliency maps are generated according to each region's strategy in the Nash equilibrium of the proposed Saliency Game. Second, we explore the complementary relationship between color and deep features and propose an Iterative Random Walk algorithm to combine saliency maps produced by the Saliency Game using different features. Iterative random walk allows sharing information across feature spaces, and detecting objects that are otherwise very hard to detect. Extensive experiments over 6 challenging datasets demonstrate the superiority of our proposed unsupervised algorithm compared to several state of the art supervised algorithms.

cs.CV↗

arXiv2017

Optimal Time Reuse in Cooperative D2D Relaying Networks

Device-to-device (D2D) communication has become one important part of the 5G cellular networks particularly due to the booming of proximity-based applications, e.g. D2D relays. However, the D2D relays may create strong interference to nearby users. Thus interference management in a cellular network with D2D relays is critical. In this paper, we study the optimal time reuse patterns in a cellular network with cooperative D2D relays and derive the corresponding optimized relaying strategies. Due to the binary association constraints and the fact that the total number of feasible time reuse patterns increase exponentially with the number of servers, we have a large scale integer programming problem which is formidable to solve. However, we show that we just need to activate a few time reuse patterns in the order of the number of users. Accordingly, a low-complexity algorithm is proposed to find the set of active time reuse patterns and solve this problem in an approximated way. Numerical simulations demonstrate that our scheme can efficiently allocate the time resources and determine the relaying strategies. Furthermore, the proposed scheme offers significant gains relative to the existing ones.

cs.NI↗

arXiv2017

Disorder-protected topological entropy after a quantum quench

Topological phases of matter are considered the bedrock of novel quantum materials as well as ideal candidates for quantum computers that possess robustness at the physical level. The robustness of the topological phase at finite temperature or away from equilibrium is therefore a very desirable feature. Disorder can improve the lifetime of the encoded topological qubits. Here we tackle the problem of the survival of the topological phase as detected by topological entropy, after a sudden quantum quench. We introduce a method to study analytically the time evolution of the system after a quantum quench and show that disorder in the couplings of the Hamiltonian of the toric code and the resulting Anderson localization can make the topological entropy resilient.

cond-mat.str-el↗

arXiv2017

Constant scalar curvature equation and the regularity of its weak solution

In this paper we study constant scalar curvature equation (CSCK), a nonlinear fourth order elliptic equation, and its weak solutions on Kähler manifolds. We first define a notion of weak solution of CSCK for an $L^\infty$ Kähler metric. The main result is to show that such a weak solution (with uniform $L^\infty$ bound) is smooth. As an application, this answers in part a conjecture of Chen regarding the regularity of $K$-energy minimizers. The new technical ingredient is a $W^{2, 2}$ regularity result for the Laplacian equation $Δ_g u=f$ on Kähler manifolds, where the metric has only $L^\infty$ coefficients. It is well-known that such a $W^{2, 2}$ regularity ($W^{2, p}$ regularity for any $p>1$) fails in general (except for dimension two) for uniform elliptic equations of the form $a^{ij}\partial^2_{ij}u=f$ for $a^{ij}\in L^\infty$, without certain smallness assumptions on the local oscillation of $a^{ij}$. We observe that the Kähler condition plays an essential role to obtain a $W^{2, 2}$ regularity for elliptic equations with only $L^\infty$ elliptic coefficients on compact manifolds.

math.DG↗