arXiv subjects

Zhi Chen

Publications and source records attributed to Zhi Chen.

At least 289 records · Page 16Linked to original sources

arXiv2019

Knowledge infused cascade convolutional neural network for segmenting retinal vessels in volumetric optical coherence tomography

We present a cascade deep neural network to segment retinal vessels in volumetric optical coherence tomography (OCT). Two types of knowledge are infused into the network for confining the searching regions. (1) Histology. The retinal vessels locate between the inner limiting membrane and the inner nuclear layer of human retina. (2) Imaging. The red blood cells inside the vessels scatter the OCT probe light forward and form projection shadows on the retinal pigment epithelium (RPE) layer, which is avascular thus perfect for localizing the retinal vessel in transverse plane. Qualitative and quantitative comparison results show that the proposed method outperforms the state-of-the-art deep learning and graph-based methods. This work demonstrates, instead of modifying the architectures of the deep networks, incorporating proper prior knowledge in the design of the image processing framework could be an efficient approach for handling such specific tasks.

eess.IV↗

arXiv2019

GLA-Net: An Attention Network with Guided Loss for Mismatch Removal

Mismatch removal is a critical prerequisite in many feature-based tasks. Recent attempts cast the mismatch removal task as a binary classification problem and solve it through deep learning based methods. In these methods, the imbalance between positive and negative classes is important, which affects network performance, i.e., Fn-score. To establish the link between Fn-score and loss, we propose to guide the loss with the Fn-score directly. We theoretically demonstrate the direct link between our Guided Loss and Fn-score during training. Moreover, we discover that outliers often impair global context in mismatch removal networks. To address this issue, we introduce the attention mechanism to mismatch removal task and propose a novel Inlier Attention Block (IA Block). To evaluate the effectiveness of our loss and IA Block, we design an end-to-end network for mismatch removal, called GLA-Net \footnote{Our code will be available in Github later.}. Experiments have shown that our network achieves the state-of-the-art performance on benchmark datasets.

cs.CV↗

arXiv2019

CANZSL: Cycle-Consistent Adversarial Networks for Zero-Shot Learning from Natural Language

Existing methods using generative adversarial approaches for Zero-Shot Learning (ZSL) aim to generate realistic visual features from class semantics by a single generative network, which is highly under-constrained. As a result, the previous methods cannot guarantee that the generated visual features can truthfully reflect the corresponding semantics. To address this issue, we propose a novel method named Cycle-consistent Adversarial Networks for Zero-Shot Learning (CANZSL). It encourages a visual feature generator to synthesize realistic visual features from semantics, and then inversely translate back synthesized the visual feature to corresponding semantic space by a semantic feature generator. Furthermore, in this paper a more challenging and practical ZSL problem is considered where the original semantics are from natural language with irrelevant words instead of clean semantics that are widely used in previous work. Specifically, a multi-modal consistent bidirectional generative adversarial network is trained to handle unseen instances by leveraging noise in the natural language. A forward one-to-many mapping from one text description to multiple visual features is coupled with an inverse many-to-one mapping from the visual space to the semantic space. Thus, a multi-modal cycle-consistency loss between the synthesized semantic representations and the ground truth can be learned and leveraged to enforce the generated semantic features to approximate to the real distribution in semantic space. Extensive experiments are conducted to demonstrate that our method consistently outperforms state-of-the-art approaches on natural language-based zero-shot learning tasks.

cs.CV↗

arXiv2019

A Unified Optimization Approach for CNN Model Inference on Integrated GPUs

Modern deep learning applications urge to push the model inference taking place at the edge devices for multiple reasons such as achieving shorter latency, relieving the burden of the network connecting to the cloud, and protecting user privacy. The Convolutional Neural Network (\emph{CNN}) is one of the most widely used model family in the applications. Given the high computational complexity of the CNN models, it is favorable to execute them on the integrated GPUs at the edge devices, which are ubiquitous and have more power and better energy efficiency than the accompanying CPUs. However, programming on integrated GPUs efficiently is challenging due to the variety of their architectures and programming interfaces. This paper proposes an end-to-end solution to execute CNN model inference on the integrated GPUs at the edge, which uses a unified IR to represent and optimize vision-specific operators on integrated GPUs from multiple vendors, as well as leverages machine learning-based scheduling search schemes to optimize computationally-intensive operators like convolution. Our solution even provides a fallback mechanism for operators not suitable or convenient to run on GPUs. The evaluation results suggest that compared to state-of-the-art solutions backed up by the vendor-provided high-performance libraries on Intel Graphics, ARM Mali GPU, and Nvidia integrated Maxwell GPU, our solution achieves similar, or even better (up to 1.62$\times$), performance on a number of popular image classification and object detection models. In addition, our solution has a wider model coverage and is more flexible to embrace new models. Our solution has been adopted in production services in AWS and is open-sourced.

cs.DC↗

arXiv2019

AgentGraph: Towards Universal Dialogue Management with Structured Deep Reinforcement Learning

Dialogue policy plays an important role in task-oriented spoken dialogue systems. It determines how to respond to users. The recently proposed deep reinforcement learning (DRL) approaches have been used for policy optimization. However, these deep models are still challenging for two reasons: 1) Many DRL-based policies are not sample-efficient. 2) Most models don't have the capability of policy transfer between different domains. In this paper, we propose a universal framework, AgentGraph, to tackle these two problems. The proposed AgentGraph is the combination of GNN-based architecture and DRL-based algorithm. It can be regarded as one of the multi-agent reinforcement learning approaches. Each agent corresponds to a node in a graph, which is defined according to the dialogue domain ontology. When making a decision, each agent can communicate with its neighbors on the graph. Under AgentGraph framework, we further propose Dual GNN-based dialogue policy, which implicitly decomposes the decision in each turn into a high-level global decision and a low-level local decision. Experiments show that AgentGraph models significantly outperform traditional reinforcement learning approaches on most of the 18 tasks of the PyDial benchmark. Moreover, when transferred from the source task to a target task, these models not only have acceptable initial performance but also converge much faster on the target task.

cs.CL↗

arXiv2018

An Incentive-Aware Job Offloading Control Framework for Mobile Edge Computing

This paper considers a scenario in which an access point (AP) is equipped with a mobile edge server of finite computing power, and serves multiple resource-hungry mobile users by charging users a price. Pricing provides users with incentives in offloading. However, existing works on pricing are based on abstract concave utility functions (e.g, the logarithm function), giving no dependence on physical layer parameters. To that end, we first introduce a novel utility function, which measures the cost reduction by offloading as compared with executing jobs locally. Based on this utility function we then formulate two offloading games, with one maximizing individual's interest and the other maximizing the overall system's interest. We analyze the structural property of the games and admit in closed form the Nash Equilibrium and the Social Equilibrium, respectively. The proposed expressions are functions of the user parameters such as the weights of computational time and energy, the distance from the AP, thus constituting an advancement over prior economic works that have considered only abstract functions. Finally, we propose an optimal pricing-based scheme, with which we prove that the interactive decision-making process with self-interested users converges to a Nash Equilibrium point equal to the Social Equilibrium point.

cs.IT↗

arXiv2018

Fast Beam Alignment for Millimeter Wave Communications: A Sparse Encoding and Phaseless Decoding Approach

In this paper, we studied the problem of beam alignment for millimeter wave (mmWave) communications, in which we assume a hybrid analog and digital beamforming structure is employed at the transmitter (i.e. base station), and an omni-directional antenna or an antenna array is used at the receiver (i.e. user). By exploiting the sparse scattering nature of mmWave channels, the beam alignment problem is formulated as a sparse encoding and phaseless decoding problem. More specifically, the problem of interest involves finding a sparse sensing matrix and an efficient recovery algorithm to recover the support and magnitude of the sparse signal from compressive phaseless measurements. A sparse bipartite graph coding (SBG-Coding) algorithm is developed for sparse encoding and phaseless decoding. Our theoretical analysis shows that, in the noiseless case, our proposed algorithm can perfectly recover the support and magnitude of the sparse signal with probability exceeding a pre-specified value from $\mathcal{O}(K^2)$ measurements, where $K$ is the number of nonzero entries of the sparse signal. The proposed algorithm has a simple decoding procedure which is computationally efficient and noise-robust. Simulation results show that our proposed method renders a reliable beam alignment in the low and moderate signal-to-noise ratio (SNR) regimes and presents a clear performance advantage over existing methods.

cs.IT↗

arXiv2018

Distributionally Robust Optimization for Sequential Decision Making

The distributionally robust Markov Decision Process (MDP) approach asks for a distributionally robust policy that achieves the maximal expected total reward under the most adversarial distribution of uncertain parameters. In this paper, we study distributionally robust MDPs where ambiguity sets for the uncertain parameters are of a format that can easily incorporate in its description the uncertainty's generalized moment as well as statistical distance information. In this way, we generalize existing works on distributionally robust MDP with generalized-moment-based and statistical-distance-based ambiguity sets to incorporate information from the former class such as moments and dispersions to the latter class that critically depends on empirical observations of the uncertain parameters. We show that, under this format of ambiguity sets, the resulting distributionally robust MDP remains tractable under mild technical conditions. To be more specific, a distributionally robust policy can be constructed by solving a sequence of one-stage convex optimization subproblems.

eess.SY↗

arXiv2018

Many Three Dimensional Objects Inspired From Finite Groups

Starting from considering deeper relationship between conjugacy classes and irreducible representations of a finite group $G$, we find some quite simple $R-$matrice defined by using finite groups. This construction produces many sets (or topological spaces) admitting braid group actions. We introduce conceptions "extended $R-$matrix" and "generalized extended $R-$matrix" generalizing Turaev's enhanced $R-$matrix, which can still give invariants of oriented links. With these new frames, we show that above $R-$matrix, together with certain commuting pairs (essentially conjugacy classes of commuting pairs ) of $G$ can give integer invariants of oriented links. We construct some group dominating these integer invariants and prove these groups are link invariant by themselves. Using the language of the (colored) tangle category, we extended above invariant to invariant of links and ribbon links colored by commuting pairs. We show given a oriented link diagram $L$, a suitable weighted sum of above invariant on all kinds of coloring of $L$ (by conjugacy classes of $G$) is invariant under both two types of Kirby moves, thus giving a invariant for closed three manifolds. We define a group dominating those invariants, and prove this group is a three manifold invariant by itself.

math.GT↗

arXiv2018

Supervised and Semi-Supervised Deep Neural Networks for CSI-Based Authentication

From the viewpoint of physical-layer authentication, spoofing attacks can be foiled by checking channel state information (CSI). Existing CSI-based authentication algorithms mostly require a deep knowledge of the channel to deliver decent performance. In this paper, we investigate CSI-based authenticators that can spare the effort to predetermine channel properties by utilizing deep neural networks (DNNs). We first propose a convolutional neural network (CNN)-enabled authenticator that is able to extract the local features in CSI. Next, we employ the recurrent neural network (RNN) to capture the dependencies between different frequencies in CSI. In addition, we propose to use the convolutional recurrent neural network (CRNN)---a combination of the CNN and the RNN---to learn local and contextual information in CSI for user authentication. To effectively train these DNNs, one needs a large amount of labeled channel records. However, it is often expensive to label large channel observations in the presence of a spoofer. In view of this, we further study a case in which only a small part of the the channel observations are labeled. To handle it, we extend these DNNs-enabled approaches into semi-supervised ones. This extension is based on a semi-supervised learning technique that employs both the labeled and unlabeled data to train a DNN. To be specific, our semi-supervised method begins by generating pseudo labels for the unlabeled channel samples through implementing the K-means algorithm in a semi-supervised manner. Subsequently, both the labeled and pseudo labeled data are exploited to pre-train a DNN, which is then fine-tuned based on the labeled channel records.

cs.LG↗

arXiv2018

Joint Power and Trajectory Design for Physical-Layer Secrecy in the UAV-Aided Mobile Relaying System

Mobile relaying is emerged as a promising technique to assist wireless communication, driven by the rapid development of unmanned aerial vehicles (UAVs). In this paper, we study secure transmission in a four-node (source, destination, mobile relay, and eavesdropper) system, wherein we focus on maximizing the secrecy rate via jointly optimizing the relay trajectory and the source/relay transmit power. Nevertheless, due to the coupling of the trajectory designing and the power allocating, the secrecy rate maximization (SRM) problem is intractable to solve. Accordingly, we propose an alternating optimization (AO) approach, wherein the trajectory designing and the power allocating are tackled in an alternating manner. Unfortunately, the trajectory designing is a nonconvex problem, and thus is still hard to solve. To circumvent the nonconvexity, we exploit sequential convex programming (SCP) to derive an iterative algorithm, which is proven to converge to a Karush-Kuhn-Tucker (KKT) point of the trajectory design problem. The simulation results demonstrate the efficacy of the joint power and trajectory design in improving the secrecy throughput.

cs.IT↗

arXiv2018

Deep Global-Connected Net With The Generalized Multi-Piecewise ReLU Activation in Deep Learning

Recent Progress has shown that exploitation of hidden layer neurons in convolution neural networks incorporating with a carefully designed activation function can yield better classification results in the field of computer vision. The paper firstly introduces a novel deep learning architecture aiming to mitigate the gradient-vanishing problem, in which the earlier hidden layer neurons could be directly connected with the last hidden layer and feed into the last layer for classification. We then design a generalized linear rectifier function as the activation function that can approximate arbitrary complex functions via training of the parameters. We will show that our design can achieve similar performance in a number of object recognition and video action benchmark tasks, under significantly less number of parameters and shallower network infrastructure, which is not only promising in training in terms of computation burden and memory usage, but is also applicable to low-computation, low-memory mobile scenarios.

cs.CV↗

arXiv2018

Intelligent Power Control for Spectrum Sharing in Cognitive Radios: A Deep Reinforcement Learning Approach

We consider the problem of spectrum sharing in a cognitive radio system consisting of a primary user and a secondary user. The primary user and the secondary user work in a non-cooperative manner. Specifically, the primary user is assumed to update its transmit power based on a pre-defined power control policy. The secondary user does not have any knowledge about the primary user's transmit power, or its power control strategy. The objective of this paper is to develop a learning-based power control method for the secondary user in order to share the common spectrum with the primary user. To assist the secondary user, a set of sensor nodes are spatially deployed to collect the received signal strength information at different locations in the wireless environment. We develop a deep reinforcement learning-based method, which the secondary user can use to intelligently adjust its transmit power such that after a few rounds of interaction with the primary user, both users can transmit their own data successfully with required qualities of service. Our experimental results show that the secondary user can interact with the primary user efficiently to reach a goal state (defined as a state in which both users can successfully transmit their data) from any initial states within a few number of steps.

cs.IT↗

arXiv2018

Partitions of the polytope of Doubly Substochastic Matrices

In this paper, we provide three different ways to partition the polytope of doubly substochastic matrices into subpolytopes via the prescribed row and column sums, the sum of all elements and the sub-defect respectively. Then we characterize the extreme points of each type of convex subpolytopes. The relations of the extreme points of the subpolytopes in the three partitions are also given.

math.CO↗

arXiv2018

Physical Layer Service Integration in 5G: Potentials and Challenges

High transmission rate and secure communication have been identified as the key targets that need to be effectively addressed by fifth generation (5G) wireless systems. In this context, the concept of physical-layer security becomes attractive, as it can establish perfect security using only the characteristics of wireless medium. Nonetheless, to further increase the spectral efficiency, an emerging concept, termed physical-layer service integration (PHY-SI), has been recognized as an effective means. Its basic idea is to combine multiple coexisting services, i.e., multicast/broadcast service and confidential service, into one integral service for one-time transmission at the transmitter side. This article first provides a tutorial on typical PHY-SI models. Furthermore, we propose some state-of-the-art solutions to improve the overall performance of PHY-SI in certain important communication scenarios. In particular, we highlight the extension of several concepts borrowed from conventional single-service communications, such as artificial noise (AN), eigenmode transmission etc., to the scenario of PHY-SI. These techniques are shown to be effective in the design of reliable and robust PHY-SI schemes. Finally, several potential research directions are identified for future work.

cs.IT↗

arXiv2017

On the maximum of the permanent of (I-A)

Let A be an n by n doubly substochastic matrix and denote σ(A) the sum of all elements of A. In this paper we give the upper bound of the permanent of (I-A) with respect to n and σ(A).

math.CO↗

arXiv2017

Adversarial Feature Matching for Text Generation

The Generative Adversarial Network (GAN) has achieved great success in generating realistic (real-valued) synthetic data. However, convergence issues and difficulties dealing with discrete data hinder the applicability of GAN to text. We propose a framework for generating realistic text via adversarial training. We employ a long short-term memory network as generator, and a convolutional network as discriminator. Instead of using the standard objective of GAN, we propose matching the high-dimensional latent feature distributions of real and synthetic sentences, via a kernelized discrepancy metric. This eases adversarial training by alleviating the mode-collapsing problem. Our experiments show superior performance in quantitative evaluation, and demonstrate that our model can generate realistic-looking sentences.

stat.ML↗

arXiv2017

Joint Power Minimization Over Multi-Carrier Two-Way Relay Networks

The study considers a three-node, two-way relaying network (TWRN) over a multi-carrier system, aiming to minimize the total power consumption of all the transmit and receiver activities. By employing digital network coding (DNC) and physical-layer network coding (PNC), respectively, as well as a novel hybrid PNC/DNC switching scheme, the total transmission power of the considered multi-carrier TWRN system is firstly analyzed; and the derived analytical expressions are then used to formulate a set of nonconvex optimization problems. We will show how those nonconvex functions are convexified for better computational tracbility. It is observed in the numerical results that, the PNC scheme generally consumes less power than DNC but becomes worse in the very low SNR regime. The proposed hybrid PNC/DNC switching scheme that takes advantage of both, is shown to outperform in all SNR regimes.

cs.IT↗