arXiv subjects

Qi Liu

Publications and source records attributed to Qi Liu.

At least 685 records · Page 38Linked to original sources

arXiv2018

Beating the classical precision limit with spin-1 Dicke state of more than 10000 atoms

Interferometry is a paradigm for most precision measurements. Using $N$ uncorrelated particles, the achievable precision for a two-mode (two-path) interferometer is bounded by the standard quantum limit (SQL), $1/\sqrt{N}$, due to the discrete (quanta) nature of individual measurements. Despite being a challenging benchmark, the two-mode SQL has been approached in a number of systems, including the LIGO and today's best atomic clocks. Employing multi-mode interferometry, the SQL becomes $1/[(M-1)\sqrt{N}]$ using M modes. Higher precision can also be achieved using entangled particles such that quantum noises from individual particles cancel out. In this work, we demonstrate an interferometric precision of $2.42^{+1.76}_{-1.29}\,$dB beyond the three-mode SQL, using balanced spin-1 (three-mode) Dicke states containing thousands of entangled atoms. The input quantum states are deterministically generated by controlled quantum phase transition and exhibit close to ideal quality. Our work shines light on the pursuit of quantum metrology beyond SQL.

cond-mat.quant-gas↗

arXiv2018

Sentence-State LSTM for Text Representation

Bi-directional LSTMs are a powerful tool for text representation. On the other hand, they have been shown to suffer various limitations due to their sequential nature. We investigate an alternative LSTM structure for encoding text, which consists of a parallel state for each word. Recurrent steps are used to perform local and global information exchange between words simultaneously, rather than incremental reading of a sequence of words. Results on various classification and sequence labelling benchmarks show that the proposed model has strong representation power, giving highly competitive performances compared to stacked BiLSTM models with similar parameter numbers.

cs.CL↗

arXiv2018

Security Analysis and Enhancement of Model Compressed Deep Learning Systems under Adversarial Attacks

DNN is presenting human-level performance for many complex intelligent tasks in real-world applications. However, it also introduces ever-increasing security concerns. For example, the emerging adversarial attacks indicate that even very small and often imperceptible adversarial input perturbations can easily mislead the cognitive function of deep learning systems (DLS). Existing DNN adversarial studies are narrowly performed on the ideal software-level DNN models with a focus on single uncertainty factor, i.e. input perturbations, however, the impact of DNN model reshaping on adversarial attacks, which is introduced by various hardware-favorable techniques such as hash-based weight compression during modern DNN hardware implementation, has never been discussed. In this work, we for the first time investigate the multi-factor adversarial attack problem in practical model optimized deep learning systems by jointly considering the DNN model-reshaping (e.g. HashNet based deep compression) and the input perturbations. We first augment adversarial example generating method dedicated to the compressed DNN models by incorporating the software-based approaches and mathematical modeled DNN reshaping. We then conduct a comprehensive robustness and vulnerability analysis of deep compressed DNN models under derived adversarial attacks. A defense technique named "gradient inhibition" is further developed to ease the generating of adversarial examples thus to effectively mitigate adversarial attacks towards both software and hardware-oriented DNNs. Simulation results show that "gradient inhibition" can decrease the average success rate of adversarial attacks from 87.99% to 4.77% (from 86.74% to 4.64%) on MNIST (CIFAR-10) benchmark with marginal accuracy degradation across various DNNs.

cs.LG↗

arXiv2018

On Modular Training of Neural Acoustics-to-Word Model for LVCSR

End-to-end (E2E) automatic speech recognition (ASR) systems directly map acoustics to words using a unified model. Previous works mostly focus on E2E training a single model which integrates acoustic and language model into a whole. Although E2E training benefits from sequence modeling and simplified decoding pipelines, large amount of transcribed acoustic data is usually required, and traditional acoustic and language modelling techniques cannot be utilized. In this paper, a novel modular training framework of E2E ASR is proposed to separately train neural acoustic and language models during training stage, while still performing end-to-end inference in decoding stage. Here, an acoustics-to-phoneme model (A2P) and a phoneme-to-word model (P2W) are trained using acoustic data and text data respectively. A phone synchronous decoding (PSD) module is inserted between A2P and P2W to reduce sequence lengths without precision loss. Finally, modules are integrated into an acousticsto-word model (A2W) and jointly optimized using acoustic data to retain the advantage of sequence modeling. Experiments on a 300- hour Switchboard task show significant improvement over the direct A2W model. The efficiency in both training and decoding also benefits from the proposed method.

cs.CL↗

arXiv2018

Probability-Scale Residuals in HIV/AIDS Research: Diagnostics and Inference

The probability-scale residual (PSR) is well defined across a wide variety of variable types and models, making it useful for studies of HIV/AIDS. In this manuscript, we highlight some of the properties of the PSR and illustrate its application with HIV data. As a residual, it can be useful for model diagnostics; we demonstrate its use with ordered categorical data and semiparametric transformation models. The PSR can also be used to construct tests of residual correlation. In fact, partial Spearman's rank correlation between $X$ and $Y$ while adjusting for covariates $Z$ can be constructed as the correlation between PSRs from models of $Y$ on $Z$ and of $X$ on $Z$. The covariance of PSRs is also useful in some settings. We apply these methods to a variety of HIV datasets including 1) a study examining risk factors for more severe forms of cervical lesions among 145 women living with HIV in Zambia, 2) a study investigating the association between 21 metabolomic biomarkers among 70 HIV-positive patients in the southeastern United States, and 3) a genome wide association study investigating the association between single nucleotide polymorphisms and tenofovir clearance among 501 HIV-positive persons participating in a multi-site randomized clinical trial.

stat.AP↗

arXiv2018

Impact of Land Use on the DOM Composition in Different Seasons in a Subtropical River Flowing through the Region Undergoing Rapid Urbanization

The dissolved organic matter (DOM) composition in river ecosystems could reflect the human impacts on the river ecosystem, and plays an important role in the carbon cycling process. We collected water and phytoplankton samples at 107 sites in the Dongjiang River in two seasons to assess the impact of the sub-catchments land use structure on the DOM composition. The results showed that (1) the forested sub-catchments had higher humic-like C1 (16.45%) and C2 (25.04%) and lower protein-like C3 (22.57%) and C4 (35.95%) than urbanized and mixed forest-agriculture sub-catchments, while the urbanized sub-catchments showed an inverse trend (4.54%, 15.51%, 33.97% and 45.98%, respectively). (2) The significant variation in the proportion of C1 and C4 between the dry and rainy seasons was recorded in both the forested and the mixed forest-agriculture sub-catchments (p<0.01), but only C4 showed an obvious seasonal variation in the urbanized sub-catchments (p<0.01). While the DOM composition was mainly related to the proportion of urbanized land and forested land year-round (p<0.01), it had stronger correlation with forested land in the dry season and urbanized land in the rainy season. (3) No significant correlation between the DOM composition and the agricultural land proportion was found in either season (p>0.05). Our findings indicated that the DOM composition was strongly dependent on the land use structure of the sub-catchments and varied seasonally, but the seasonal variation pattern could be disturbed by human activities in the extensively urbanized catchments. Notably, the higher C4 proportion compared with those in temperate rivers implied a larger amount of CO2 released from the subtropical rivers into the atmosphere when considering bioavailability.

q-bio.PE↗

arXiv2018

Divide and Recombine for Large and Complex Data: Model Likelihood Functions using MCMC

In Divide & Recombine (D&R), big data are divided into subsets, each analytic method is applied to subsets, and the outputs are recombined. This enables deep analysis and practical computational performance. An innovate D\&R procedure is proposed to compute likelihood functions of data-model (DM) parameters for big data. The likelihood-model (LM) is a parametric probability density function of the DM parameters. The density parameters are estimated by fitting the density to MCMC draws from each subset DM likelihood function, and then the fitted densities are recombined. The procedure is illustrated using normal and skew-normal LMs for the logistic regression DM.

stat.ME↗

arXiv2018

Robust formation of topological Hall effect in MnGa/heavy metal bilayers

We have investigated the topological Hall effect (THE) in MnGa/Pt and MnGa/Ta bilayers induced by interfacial Dzyaloshinskii-Moriya interaction (DMI). The most evident THE signals have been found based on the MnGa films with small critical DMI energy constant Dc. The large topological portion of the Hall signal from the total Hall signal has been extracted in the whole temperature range from 5 to 300 K. These results open up the exploration of the DMI induced magnetic behavior based on the bulk perpendicular magnetic anisotropy materials for fundamental physics and magnetic storage technologies.

cond-mat.mtrl-sci↗

arXiv2017

SHINE: Signed Heterogeneous Information Network Embedding for Sentiment Link Prediction

In online social networks people often express attitudes towards others, which forms massive sentiment links among users. Predicting the sign of sentiment links is a fundamental task in many areas such as personal advertising and public opinion analysis. Previous works mainly focus on textual sentiment classification, however, text information can only disclose the "tip of the iceberg" about users' true opinions, of which the most are unobserved but implied by other sources of information such as social relation and users' profile. To address this problem, in this paper we investigate how to predict possibly existing sentiment links in the presence of heterogeneous information. First, due to the lack of explicit sentiment links in mainstream social networks, we establish a labeled heterogeneous sentiment dataset which consists of users' sentiment relation, social relation and profile knowledge by entity-level sentiment extraction method. Then we propose a novel and flexible end-to-end Signed Heterogeneous Information Network Embedding (SHINE) framework to extract users' latent representations from heterogeneous networks and predict the sign of unobserved sentiment links. SHINE utilizes multiple deep autoencoders to map each user into a low-dimension feature space while preserving the network structure. We demonstrate the superiority of SHINE over state-of-the-art baselines on link prediction and node recommendation in two real-world datasets. The experimental results also prove the efficacy of SHINE in cold start scenario.

stat.ML↗

arXiv2017

Interference coupling analysis based on a hybrid method: application for radio telescope system

Working in a way that passively receives the electromagnetic radiation of celestial body, the radio telescope is easily disturbed by external radio frequency interference (RFI) as well as the electromagnetic interference (EMI) generated by electric and electronic components operating at telescope site. The quantitative analysis for these interferences must be taken into account carefully for further electromagnetic protection of the radio telescope. In this paper, based on the electromagnetic topology (EMT) theory, a hybrid method that combines Baum-Liu-Tesche (BLT) equation and transfer function is proposed. In this method, the coupling path of radio telescope is divided into strong coupling and weak coupling sub-paths, and the coupling intensity criterion is proposed by analyzing the conditions that BLT equation simplifies to transfer function. According to the coupling intensity criterion, the topological model of a typical radio telescope system is established. The proposed method is used to solve the interference response of the radio telescope system by analyzing the subsystems with different coupling modes respectively and then integrating the responses of the subsystems as the response of the entire system. The validity of the proposed method is verified numerically. The results indicate that the proposed method, compared with the direct solving method, reduces the difficulty and improves the efficiency of the interference prediction.

astro-ph.IM↗

arXiv2017

Applying High-Resolution Visible Imagery to Satellite Melt Pond Fraction Retrieval: A Neural Network Approach

During summer, melt ponds have a significant influence on Arctic sea-ice albedo. The melt pond fraction (MPF) also has the ability to forecast the Arctic sea-ice in a certain period. It is important to retrieve accurate melt pond fraction (MPF) from satellite data for Arctic research. This paper proposes a satellite MPF retrieval model based on the multi-layer neural network, named MPF-NN. Our model uses multi-spectral satellite data as model input and MPF information from multi-site and multi-period visible imagery as prior knowledge for modeling. It can effectively model melt ponds evolution of different regions and periods over the Arctic. Evaluation results show that the MPF retrieved from MODIS data using the proposed model has an RMSE of 3.91% and a correlation coefficient of 0.73. The seasonal distribution of MPF is also consistent with previous results.

physics.ao-ph↗

arXiv2017

I Ate This: A Photo-based Food Journaling System with Expert Feedback

What we eat is one of the most frequent and important health decisions we make in daily life, yet it remains notoriously difficult to capture and understand. Effective food journaling is thus a grand challenge in personal health informatics. In this paper we describe a system for food journaling called I Ate This, which is inspired by the Remote Food Photography Method (RFPM). I Ate This is simple: you use a smartphone app to take a photo and give a very basic description of any food or beverage you are about to consume. Later, a qualified dietitian will evaluate your photo, giving you feedback on how you did and where you can improve. The aim of I Ate This is to provide a convenient, visual and reliable way to help users learn from their eating habits and nudge them towards better choices each and every day. Ultimately, this incremental approach can lead to long-term behaviour change. Our goal is to bring RFPM to a wider audience, through APIs that can be incorporated into other apps.

cs.HC↗

arXiv2017

Deterministic entanglement generation from driving through quantum phase transitions

Many-body entanglement is often created through system evolution, aided by non-linear interactions between the constituting particles. The very dynamics, however, can also lead to fluctuations and degradation of the entanglement if the interactions cannot be controlled. Here, we demonstrate near-deterministic generation of an entangled twin-Fock condensate of $\sim11000$ atoms by driving a $^{87}$Rb Bose-Einstein condensate undergoing spin mixing through two consecutive quantum phase transitions (QPTs). We directly observe number squeezing of $10.7\pm0.6$ dB and normalized collective spin length of $0.99\pm0.01$. Together, these observations allow us to infer an entanglement-enhanced phase sensitivity of $\sim6$ dB beyond the standard quantum limit and an entanglement breadth of $\sim910$ atoms. Our work highlights the power of generating large-scale useful entanglement by taking advantage of the different entanglement landscapes separated by QPTs.

cond-mat.quant-gas↗

arXiv2016

Cryptanalyzing image encryption using chaotic logistic map

Chaotic behavior arises from very simple non-linear dynamical equation of logistic map which makes it was used often in designing chaotic image encryption schemes. However, some properties of chaotic maps can also facilitate cryptanalysis especially when they are implemented in digital domain. Utilizing stable distribution of the chaotic states generated by iterating the logistic map, this paper presents a typical example to show insecurity of an image encryption scheme using chaotic logistic map. This work will push encryption and chaos be combined in a more effective way.

cs.CR↗

arXiv2016

Modeling the time-periodicity of in-degree distributions in scientific citation networks

In a range of citation networks, the in-degree distributions boast time-periodicity---the distributions of citations per article published each year present similar scale-free tails. This phenomenon can be regarded as a consequence of the emergence of hot topics and the existence of the "burst" phenomenon. With this inference considered, a geometric model based on our previous study is established, in which the sizes of the influence zones of nodes follow the same power-law distribution and decrease with their ages. The model successfully reproduces the time-periodicity of the in-degree distributions of the empirical data, and accounts for the presence of citation burst as well. Moreover, a reasonable explanation for the emergence of the scale-free tails by regarding the citation behavior between articles as a "yes/no" experiment is presented. The model can also predict the time-periodicity of the local clustering coefficients, which indicates that the model is a good tool in researches on the evolutionary mechanism of citation networks.

physics.soc-ph↗

arXiv2016

Privacy Preserving Linear Programming

With the rapid increase in computing, storage and networking resources, data is not only collected and stored, but also analyzed. This creates a serious privacy problem which often inhibits the use of this data. In this chapter, we investigate and resolve the privacy issues in a fundamental optimization problem -- linear programming (LP) which is formulated by data collected from different parties. We first consider the case where the objective function and constraints of the linear programming problem are not partitioned between two parties where one party privately holds the objective function while the other party privately holds the constraints. Second, we present a privacy preserving technique for the case that objective function and constraints are arbitrarily partitioned between two parties where each party privately holds a share of objective function and constraints. Finally, we extend the technique for securely solving two-party arbitrarily partitioned linear programming problems to a multi-party scenario. In summary, we propose a set of efficient and secure transformation based techniques that create significant value-added benefits of being independent of the specific algorithms used for solving the linear programming problem.

cs.CR↗

arXiv2016

Dynamical Gauge Effects and Holographic Scaling of Non-Equilibrium Motion in a Disordered and Dissipative Atomic Gas

We present a table-top realization of a non-equilibrium quantum system described by a dynamical gauge field propagating on an effectively curved space and time manifold. The system is formed by neutral atoms interacting with both a conservative disordered optical field and a dissipative pumping field. In the presence of a sufficiently dark state, we demonstrate non-equilibrium behavior reminiscent of the information paradox in black hole physics. At a well-defined transition point, the analog of gauge-boson mass is seen to vanish, inducing scale-invariant behavior as a Higgs-like mechanism is removed. The subsequent scaling behavior can be understood using the holographic principle with a tunable analog of the Planck length derived from the scaling of disorder. These effects suggest a range of new phenomena in weakly dissipative quantum systems, including the presence of analog forms of emergent gravitation.

cond-mat.quant-gas↗

arXiv2016

A geometric graph model of citation networks with linearly growing node-increment

Due to the fact that the numbers of annually published papers have witnessed a linear growth in some citation networks, a geometric model is thus proposed to predict some statistical features of those networks, in which the academic influence scopes of the papers are denoted through specific geometric areas related to time and space. In the model, nodes (papers) are uniformly and randomly sprinkled onto a cluster of circles of the Minkowski space whose centers are on the time axis. Edges (citations) are linked according to an influence mechanism which indicates that an existing paper will be cited by a new paper located in its influence zone. Considering the citations among papers in different disciplines, an interdisciplinary citation mechanism is added to the model in which some papers with a small probability of being chosen will cite some existing papers randomly and uniformly. Different from most existing models that only study the power-law tail of the in-degree distribution, this model also characterizes the overall in-degree distribution. Moreover, it presents the description of some other important statistical characteristics of real networks, such as in- and out-assortativity, giant component and clear community structure. Therefore, it is reasonable to believe that a good example is provided in the paper to study real networks by geometric graphs.

physics.soc-ph↗