arXiv subjects

Jun Guo

Publications and source records attributed to Jun Guo.

At least 145 records · Page 8Linked to original sources

arXiv2018

Constrained Sparse Subspace Clustering with Side-Information

Subspace clustering refers to the problem of segmenting high dimensional data drawn from a union of subspaces into the respective subspaces. In some applications, partial side-information to indicate "must-link" or "cannot-link" in clustering is available. This leads to the task of subspace clustering with side-information. However, in prior work the supervision value of the side-information for subspace clustering has not been fully exploited. To this end, in this paper, we present an enhanced approach for constrained subspace clustering with side-information, termed Constrained Sparse Subspace Clustering plus (CSSC+), in which the side-information is used not only in the stage of learning an affinity matrix but also in the stage of spectral clustering. Moreover, we propose to estimate clustering accuracy based on the partial side-information and theoretically justify the connection to the ground-truth clustering accuracy in terms of the Rand index. We conduct experiments on three cancer gene expression datasets to validate the effectiveness of our proposals.

cs.CV↗

arXiv2018

SketchMate: Deep Hashing for Million-Scale Human Sketch Retrieval

We propose a deep hashing framework for sketch retrieval that, for the first time, works on a multi-million scale human sketch dataset. Leveraging on this large dataset, we explore a few sketch-specific traits that were otherwise under-studied in prior literature. Instead of following the conventional sketch recognition task, we introduce the novel problem of sketch hashing retrieval which is not only more challenging, but also offers a better testbed for large-scale sketch analysis, since: (i) more fine-grained sketch feature learning is required to accommodate the large variations in style and abstraction, and (ii) a compact binary code needs to be learned at the same time to enable efficient retrieval. Key to our network design is the embedding of unique characteristics of human sketch, where (i) a two-branch CNN-RNN architecture is adapted to explore the temporal ordering of strokes, and (ii) a novel hashing loss is specifically designed to accommodate both the temporal and abstract traits of sketches. By working with a 3.8M sketch dataset, we show that state-of-the-art hashing models specifically engineered for static images fail to perform well on temporal sketch data. Our network on the other hand not only offers the best retrieval performance on various code sizes, but also yields the best generalization performance under a zero-shot setting and when re-purposed for sketch recognition. Such superior performances effectively demonstrate the benefit of our sketch-specific design.

cs.CV↗

arXiv2018

Robust Actor-Critic Contextual Bandit for Mobile Health (mHealth) Interventions

We consider the actor-critic contextual bandit for the mobile health (mHealth) intervention. State-of-the-art decision-making algorithms generally ignore the outliers in the dataset. In this paper, we propose a novel robust contextual bandit method for the mHealth. It can achieve the conflicting goal of reducing the influence of outliers while seeking for a similar solution compared with the state-of-the-art contextual bandit methods on the datasets without outliers. Such performance relies on two technologies: (1) the capped-$\ell_{2}$ norm; (2) a reliable method to set the thresholding hyper-parameter, which is inspired by one of the most fundamental techniques in the statistics. Although the model is non-convex and non-differentiable, we propose an effective reweighted algorithm and provide solid theoretical analyses. We prove that the proposed algorithm can find sufficiently decreasing points after each iteration and finally converges after a finite number of iterations. Extensive experiment results on two datasets demonstrate that our method can achieve almost identical results compared with state-of-the-art contextual bandit methods on the dataset without outliers, and significantly outperform those state-of-the-art methods on the badly noised dataset with outliers in a variety of parameter settings.

cs.LG↗

arXiv2017

Inference-Based Similarity Search in Randomized Montgomery Domains for Privacy-Preserving Biometric Identification

Similarity search is essential to many important applications and often involves searching at scale on high-dimensional data based on their similarity to a query. In biometric applications, recent vulnerability studies have shown that adversarial machine learning can compromise biometric recognition systems by exploiting the biometric similarity information. Existing methods for biometric privacy protection are in general based on pairwise matching of secured biometric templates and have inherent limitations in search efficiency and scalability. In this paper, we propose an inference-based framework for privacy-preserving similarity search in Hamming space. Our approach builds on an obfuscated distance measure that can conceal Hamming distance in a dynamic interval. Such a mechanism enables us to systematically design statistically reliable methods for retrieving most likely candidates without knowing the exact distance values. We further propose to apply Montgomery multiplication for generating search indexes that can withstand adversarial similarity analysis, and show that information leakage in randomized Montgomery domains can be made negligibly small. Our experiments on public biometric datasets demonstrate that the inference-based approach can achieve a search accuracy close to the best performance possible with secure computation methods, but the associated cost is reduced by orders of magnitude compared to cryptographic primitives.

cs.CR↗

arXiv2017

Movement-efficient Sensor Deployment in Wireless Sensor Networks

We study a mobile wireless sensor network (MWSN) consisting of multiple mobile sensors or robots. Two key issues in MWSNs - energy consumption, which is dominated by sensor movement, and sensing coverage - have attracted plenty of attention, but the interaction of these issues is not well studied. To take both sensing coverage and movement energy consumption into consideration, we model the sensor deployment problem as a constrained source coding problem. %, which can be applied to different coverage tasks, such as area coverage, target coverage, and barrier coverage. Our goal is to find an optimal sensor deployment to maximize the sensing coverage with specific energy constraints. We derive necessary conditions to the optimal sensor deployment with (i) total energy constraint and (ii) network lifetime constraint. Using these necessary conditions, we design Lloyd-like algorithms to provide a trade-off between sensing coverage and energy consumption. Simulation results show that our algorithms outperform the existing relocation algorithms.

cs.IT↗

arXiv2017

Group-driven Reinforcement Learning for Personalized mHealth Intervention

Due to the popularity of smartphones and wearable devices nowadays, mobile health (mHealth) technologies are promising to bring positive and wide impacts on people's health. State-of-the-art decision-making methods for mHealth rely on some ideal assumptions. Those methods either assume that the users are completely homogenous or completely heterogeneous. However, in reality, a user might be similar with some, but not all, users. In this paper, we propose a novel group-driven reinforcement learning method for the mHealth. We aim to understand how to share information among similar users to better convert the limited user information into sharper learned RL policies. Specifically, we employ the K-means clustering method to group users based on their trajectory information similarity and learn a shared RL policy for each group. Extensive experiment results have shown that our method can achieve clear gains over the state-of-the-art RL methods for mHealth.

cs.LG↗

arXiv2017

Adversarial Network Bottleneck Features for Noise Robust Speaker Verification

In this paper, we propose a noise robust bottleneck feature representation which is generated by an adversarial network (AN). The AN includes two cascade connected networks, an encoding network (EN) and a discriminative network (DN). Mel-frequency cepstral coefficients (MFCCs) of clean and noisy speech are used as input to the EN and the output of the EN is used as the noise robust feature. The EN and DN are trained in turn, namely, when training the DN, noise types are selected as the training labels and when training the EN, all labels are set as the same, i.e., the clean speech label, which aims to make the AN features invariant to noise and thus achieve noise robustness. We evaluate the performance of the proposed feature on a Gaussian Mixture Model-Universal Background Model based speaker verification system, and make comparison to MFCC features of speech enhanced by short-time spectral amplitude minimum mean square error (STSA-MMSE) and deep neural network-based speech enhancement (DNN-SE) methods. Experimental results on the RSR2015 database show that the proposed AN bottleneck feature (AN-BN) dramatically outperforms the STSA-MMSE and DNN-SE based MFCCs for different noise types and signal-to-noise ratios. Furthermore, the AN-BN feature is able to improve the speaker verification performance under the clean condition.

cs.SD↗

arXiv2017

Decorrelation of Neutral Vector Variables: Theory and Applications

In this paper, we propose novel strategies for neutral vector variable decorrelation. Two fundamental invertible transformations, namely serial nonlinear transformation and parallel nonlinear transformation, are proposed to carry out the decorrelation. For a neutral vector variable, which is not multivariate Gaussian distributed, the conventional principal component analysis (PCA) cannot yield mutually independent scalar variables. With the two proposed transformations, a highly negatively correlated neutral vector can be transformed to a set of mutually independent scalar variables with the same degrees of freedom. We also evaluate the decorrelation performances for the vectors generated from a single Dirichlet distribution and a mixture of Dirichlet distributions. The mutual independence is verified with the distance correlation measurement. The advantages of the proposed decorrelation strategies are intensively studied and demonstrated with synthesized data and practical application evaluations.

cs.CV↗

arXiv2017

The Role of Data Analysis in the Development of Intelligent Energy Networks

Data analysis plays an important role in the development of intelligent energy networks (IENs). This article reviews and discusses the application of data analysis methods for energy big data. The installation of smart energy meters has provided a huge volume of data at different time resolutions, suggesting data analysis is required for clustering, demand forecasting, energy generation optimization, energy pricing, monitoring and diagnostics. The currently adopted data analysis technologies for IENs include pattern recognition, machine learning, data mining, statistics methods, etc. However, existing methods for data analysis cannot fully meet the requirements for processing the big data produced by the IENs and, therefore, more comprehensive data analysis methods are needed to handle the increasing amount of data and to mine more valuable information.

cs.OH↗

arXiv2017

Cross-modal Subspace Learning for Fine-grained Sketch-based Image Retrieval

Sketch-based image retrieval (SBIR) is challenging due to the inherent domain-gap between sketch and photo. Compared with pixel-perfect depictions of photos, sketches are iconic renderings of the real world with highly abstract. Therefore, matching sketch and photo directly using low-level visual clues are unsufficient, since a common low-level subspace that traverses semantically across the two modalities is non-trivial to establish. Most existing SBIR studies do not directly tackle this cross-modal problem. This naturally motivates us to explore the effectiveness of cross-modal retrieval methods in SBIR, which have been applied in the image-text matching successfully. In this paper, we introduce and compare a series of state-of-the-art cross-modal subspace learning methods and benchmark them on two recently released fine-grained SBIR datasets. Through thorough examination of the experimental results, we have demonstrated that the subspace learning can effectively model the sketch-photo domain-gap. In addition we draw a few key insights to drive future research.

cs.CV↗

arXiv2017

One-to-Many Network for Visually Pleasing Compression Artifacts Reduction

We consider the compression artifacts reduction problem, where a compressed image is transformed into an artifact-free image. Recent approaches for this problem typically train a one-to-one mapping using a per-pixel $L_2$ loss between the outputs and the ground-truths. We point out that these approaches used to produce overly smooth results, and PSNR doesn't reflect their real performance. In this paper, we propose a one-to-many network, which measures output quality using a perceptual loss, a naturalness loss, and a JPEG loss. We also avoid grid-like artifacts during deconvolution using a "shift-and-average" strategy. Extensive experimental results demonstrate the dramatic visual improvement of our approach over the state of the arts.

cs.CV↗

arXiv2017

Energy Efficiency in Two-Tiered Wireless Sensor Networks

We study a two-tiered wireless sensor network (WSN) consisting of $N$ access points (APs) and $M$ base stations (BSs). The sensing data, which is distributed on the sensing field according to a density function $f$, is first transmitted to the APs and then forwarded to the BSs. Our goal is to find an optimal deployment of APs and BSs to minimize the average weighted total, or Lagrangian, of sensor and AP powers. For $M=1$, we show that the optimal deployment of APs is simply a linear transformation of the optimal $N$-level quantizer for density $f$, and the sole BS should be located at the geometric centroid of the sensing field. Also, for a one-dimensional network and uniform $f$, we determine the optimal deployment of APs and BSs for any $N$ and $M$. Moreover, to numerically optimize node deployment for general scenarios, we propose one- and two-tiered Lloyd algorithms and analyze their convergence properties. Simulation results show that, when compared to random deployment, our algorithms can save up to 79\% of the power on average.

cs.NI↗

arXiv2017

DNN Filter Bank Cepstral Coefficients for Spoofing Detection

With the development of speech synthesis techniques, automatic speaker verification systems face the serious challenge of spoofing attack. In order to improve the reliability of speaker verification systems, we develop a new filter bank based cepstral feature, deep neural network filter bank cepstral coefficients (DNN-FBCC), to distinguish between natural and spoofed speech. The deep neural network filter bank is automatically generated by training a filter bank neural network (FBNN) using natural and synthetic speech. By adding restrictions on the training rules, the learned weight matrix of FBNN is band-limited and sorted by frequency, similar to the normal filter bank. Unlike the manually designed filter bank, the learned filter bank has different filter shapes in different channels, which can capture the differences between natural and synthetic speech more effectively. The experimental results on the ASVspoof {2015} database show that the Gaussian mixture model maximum-likelihood (GMM-ML) classifier trained by the new feature performs better than the state-of-the-art linear frequency cepstral coefficients (LFCC) based classifier, especially on detecting unknown attacks.

cs.SD↗

arXiv2017

Higher Order Mode Entanglement in a Type II Optical Parametric Oscillator

Nonclassical beams in high order spatial modes have attracted much interest but they exhibit much less squeezing and entanglement than the fundamental spatial modes, limiting their applications. We experimentally demonstrate the relation between pump modes and entanglement of first-order HG modes (HG10 entangled states) in a type II OPO and show that the maximum entanglement of high order spatial modes can be obtained by optimizing the pump spatial mode. To our knowledge, this is the first time to report this. Utilizing the optimal pump mode, the HG10 mode threshold can be reached easily without HG00 oscillation and HG10 entanglement is enhanced by 53.5% over HG00 pumping. The technique is broadly applicable to entanglement generation in high order modes.

quant-ph↗

arXiv2016

A comprehensive approach to dark matter studies: exploration of simplified top-philic models

Studies of dark matter lie at the interface of collider physics, astrophysics and cosmology. Constraining models featuring dark matter candidates entails the capability to provide accurate predictions for large sets of observables and compare them to a wide spectrum of data. We present a framework which, starting from a model lagrangian, allows one to consistently and systematically make predictions, as well as to confront those predictions with a multitude of experimental results. As an application, we consider a class of simplified dark matter models where a scalar mediator couples only to the top quark and a fermionic dark sector (i.e. the simplified top-philic dark matter model). We study in detail the complementarity of relic density, direct/indirect detection and collider searches in constraining the multi-dimensional model parameter space, and efficiently identify regions where individual approaches to dark matter detection provide the most stringent bounds. In the context of collider studies of dark matter, we point out the complementarity of LHC searches in probing different regions of the model parameter space with final states involving top quarks, photons, jets and/or missing energy. Our study of dark matter production at the LHC goes beyond the tree-level approximation and we show examples of how higher-order corrections to dark matter production processes can affect the interpretation of the experimental results.

hep-ph↗

arXiv2016

Investigating light NMSSM pseudoscalar states with boosted ditau tagging

We study a class of realizations of the Next-to-Minimal Supersymmetric Standard Model that is motivated by dark matter and Higgs data, and in which the lightest pseudoscalar Higgs boson mass is smaller than twice the bottom quark mass and greater than twice the tau lepton mass. In such scenarios, the lightest pseudoscalar Higgs boson can be copiously produced at the LHC from the decay of heavier superpartners and will dominantly further decay into a pair of tau leptons that is generally boosted. We make use of a boosted object tagging technique designed to tag such a ditau jet, and estimate the sensitivity of the LHC to the considered supersymmetric scenarios with 20 to 50~fb$^{-1}$ of proton-proton collisions at a center-of-mass energy of 13~TeV.

hep-ph↗

arXiv2016

Sensor Deployment with Limited Communication Range in Homogeneous and Heterogeneous Wireless Sensor Networks

We study the heterogeneous wireless sensor networks (WSNs) and propose the necessary condition of the optimal sensor deployment. Similar to that in homogeneous WSNs, the necessary condition implies that every sensor node location should coincide with the centroid of its own optimal sensing region. Moreover, we discuss the dynamic sensor deployment in both homogeneous and heterogeneous WSNs with limited communication range for the sensor nodes. The purpose of sensor deployment is to improve sensing performance, reflected by distortion and coverage. We model the sensor deployment problem as a source coding problem with distortion reflecting sensing accuracy. Traditionally, coverage is the area covered by the sensor nodes. However, when the communication range is limited, a WSN may be divided into several disconnected sub-graphs. Under such a scenario, neither the conventional distortion nor the coverage represents the sensing performance as the collected data in disconnected sub-graphs cannot be communicated with the access point. By defining an appropriate distortion measure, we propose a Restrained Lloyd (RL) algorithm and a Deterministic Annealing (DA) algorithm to optimize sensor deployment in both homogeneous and heterogeneous WSNs. Our simulation results show that both DA and RL algorithms outperform the existing Lloyd algorithm when communication range is limited.

cs.NI↗

arXiv2016

Effective Clipart Image Vectorization Through Direct Optimization of Bezigons

Bezigons, i.e., closed paths composed of Bézier curves, have been widely employed to describe shapes in image vectorization results. However, most existing vectorization techniques infer the bezigons by simply approximating an intermediate vector representation (such as polygons). Consequently, the resultant bezigons are sometimes imperfect due to accumulated errors, fitting ambiguities, and a lack of curve priors, especially for low-resolution images. In this paper, we describe a novel method for vectorizing clipart images. In contrast to previous methods, we directly optimize the bezigons rather than using other intermediate representations; therefore, the resultant bezigons are not only of higher fidelity compared with the original raster image but also more reasonable because they were traced by a proficient expert. To enable such optimization, we have overcome several challenges and have devised a differentiable data energy as well as several curve-based prior terms. To improve the efficiency of the optimization, we also take advantage of the local control property of bezigons and adopt an overlapped piecewise optimization strategy. The experimental results show that our method outperforms both the current state-of-the-art method and commonly used commercial software in terms of bezigon quality.

cs.GR↗