arXiv subjects

Zhi Chen

Publications and source records attributed to Zhi Chen.

At least 271 records · Page 15Linked to original sources

arXiv2020

Deep Reinforcement Learning for On-line Dialogue State Tracking

Dialogue state tracking (DST) is a crucial module in dialogue management. It is usually cast as a supervised training problem, which is not convenient for on-line optimization. In this paper, a novel companion teaching based deep reinforcement learning (DRL) framework for on-line DST optimization is proposed. To the best of our knowledge, this is the first effort to optimize the DST module within DRL framework for on-line task-oriented spoken dialogue systems. In addition, dialogue policy can be further jointly updated. Experiments show that on-line DST optimization can effectively improve the dialogue manager performance while keeping the flexibility of using predefined policy. Joint training of both DST and policy can further improve the performance.

cs.CL↗

arXiv2020

Structured Hierarchical Dialogue Policy with Graph Neural Networks

Dialogue policy training for composite tasks, such as restaurant reservation in multiple places, is a practically important and challenging problem. Recently, hierarchical deep reinforcement learning (HDRL) methods have achieved good performance in composite tasks. However, in vanilla HDRL, both top-level and low-level policies are all represented by multi-layer perceptrons (MLPs) which take the concatenation of all observations from the environment as the input for predicting actions. Thus, traditional HDRL approach often suffers from low sampling efficiency and poor transferability. In this paper, we address these problems by utilizing the flexibility of graph neural networks (GNNs). A novel ComNet is proposed to model the structure of a hierarchical agent. The performance of ComNet is tested on composited tasks of the PyDial benchmark. Experiments show that ComNet outperforms vanilla HDRL systems with performance close to the upper bound. It not only achieves sample efficiency but also is more robust to noise while maintaining the transferability to other composite tasks.

cs.CL↗

arXiv2020

Dual Learning for Dialogue State Tracking

In task-oriented multi-turn dialogue systems, dialogue state refers to a compact representation of the user goal in the context of dialogue history. Dialogue state tracking (DST) is to estimate the dialogue state at each turn. Due to the dependency on complicated dialogue history contexts, DST data annotation is more expensive than single-sentence language understanding, which makes the task more challenging. In this work, we formulate DST as a sequence generation problem and propose a novel dual-learning framework to make full use of unlabeled data. In the dual-learning framework, there are two agents: the primal tracker agent (utterance-to-state generator) and the dual utterance generator agent (state-to-utterance genera-tor). Compared with traditional supervised learning framework, dual learning can iteratively update both agents through the reconstruction error and reward signal respectively without labeled data. Reward sparsity problem is hard to solve in previous DST methods. In this work, the reformulation of DST as a sequence generation model effectively alleviates this problem. We call this primal tracker agent dual-DST. Experimental results on MultiWOZ2.1 dataset show that the proposed dual-DST works very well, especially when labelled data is limited. It achieves comparable performance to the system where labeled data is fully used.

cs.CL↗

arXiv2020

CREDIT: Coarse-to-Fine Sequence Generation for Dialogue State Tracking

In dialogue systems, a dialogue state tracker aims to accurately find a compact representation of the current dialogue status, based on the entire dialogue history. While previous approaches often define dialogue states as a combination of separate triples ({\em domain-slot-value}), in this paper, we employ a structured state representation and cast dialogue state tracking as a sequence generation problem. Based on this new formulation, we propose a {\bf C}oa{\bf R}s{\bf E}-to-fine {\bf DI}alogue state {\bf T}racking ({\bf CREDIT}) approach. Taking advantage of the structured state representation, which is a marked language sequence, we can further fine-tune the pre-trained model (by supervised learning) by optimizing natural language metrics with the policy gradient method. Like all generative state tracking methods, CREDIT does not rely on pre-defined dialogue ontology enumerating all possible slot values. Experiments demonstrate our tracker achieves encouraging joint goal accuracy for the five domains in MultiWOZ 2.0 and MultiWOZ 2.1 datasets.

cs.CL↗

arXiv2020

Distributed Structured Actor-Critic Reinforcement Learning for Universal Dialogue Management

The task-oriented spoken dialogue system (SDS) aims to assist a human user in accomplishing a specific task (e.g., hotel booking). The dialogue management is a core part of SDS. There are two main missions in dialogue management: dialogue belief state tracking (summarising conversation history) and dialogue decision-making (deciding how to reply to the user). In this work, we only focus on devising a policy that chooses which dialogue action to respond to the user. The sequential system decision-making process can be abstracted into a partially observable Markov decision process (POMDP). Under this framework, reinforcement learning approaches can be used for automated policy optimization. In the past few years, there are many deep reinforcement learning (DRL) algorithms, which use neural networks (NN) as function approximators, investigated for dialogue policy.

cs.CL↗

arXiv2020

Stochastic Modeling Approaches for Analyzing Blockchain: A Survey

Blockchain technology has been attracting much attention from both academia and industry. It brings many benefits to various applications like Internet of Things. However, there are critical issues to be addressed before its widespread deployment, such as transaction efficiency, bandwidth bottleneck, and security. Techniques are being explored to tackle these issues. Stochastic modeling, as one of these techniques, has been applied to analyze a variety of blockchain characteristics, but there is a lack of a comprehensive survey on it. In this survey, we aim to fill the gap and review the stochastic models proposed to address common issues in blockchain. Firstly, this paper provides the basic knowledge of blockchain technology and stochastic models. Then, according to different objects, the stochastic models for blockchain analysis are divided into network-oriented and application-oriented (mainly refer to cryptocurrency). The network-oriented stochastic models are further classified into two categories, namely, performance and security. About the application-oriented stochastic models, the widest adoption mainly concentrates on the price prediction of cryptocurrency. Moreover, we provide analysis and comparison in detail on every taxonomy and discuss the strengths and weaknesses of the related works to serve guides for further researches. Finally, challenges and future research directions are given to apply stochastic modeling approaches to study blockchain. By analyzing and classifying the existing researches, we hope that our survey can provide suggestions for the researchers who are interested in blockchain and good at using stochastic models as a tool to address problems.

cs.CR↗

arXiv2020

Optimal Pricing for Job Offloading in the MEC System with Two Priority Classes

Multi-Access edge computing (MEC) is an emerging paradigm where users offload computationally intensive jobs to the Access Point (AP). Given that the AP's resources are shared by selfish users, pricing is a useful tool for incentivising users to internalize the negative externality of delay they cause to other users. Nevertheless, different users have different negative valuations towards delay as some are more delay sensitive. To serve heterogeneous users, we propose a priority pricing scheme where users can get served first for a higher price. Our goal is to find the prices such that in decision making, users will choose the class and the offloading frequency that jointly maximize social welfare. With the assumption that the AP knows users' profit functions, we derive in semi-closed form the optimal prices. However in practice, the reporting of users's profit information incurs a large signalling overhead. Besides, in reality users might falsely report their private profit information. To overcome this, we further propose a learning-based pricing mechanism where no knowledge of individual user profit functions is required. At equilibrium, the optimal prices and average edge delays are learnt, and users have chosen the correct priority class and offload at the socially optimal frequency.

cs.IT↗

arXiv2020

Rethinking Generative Zero-Shot Learning: An Ensemble Learning Perspective for Recognising Visual Patches

Zero-shot learning (ZSL) is commonly used to address the very pervasive problem of predicting unseen classes in fine-grained image classification and other tasks. One family of solutions is to learn synthesised unseen visual samples produced by generative models from auxiliary semantic information, such as natural language descriptions. However, for most of these models, performance suffers from noise in the form of irrelevant image backgrounds. Further, most methods do not allocate a calculated weight to each semantic patch. Yet, in the real world, the discriminative power of features can be quantified and directly leveraged to improve accuracy and reduce computational complexity. To address these issues, we propose a novel framework called multi-patch generative adversarial nets (MPGAN) that synthesises local patch features and labels unseen classes with a novel weighted voting strategy. The process begins by generating discriminative visual features from noisy text descriptions for a set of predefined local patches using multiple specialist generative models. The features synthesised from each patch for unseen classes are then used to construct an ensemble of diverse supervised classifiers, each corresponding to one local patch. A voting strategy averages the probability distributions output from the classifiers and, given that some patches are more discriminative than others, a discrimination-based attention mechanism helps to weight each patch accordingly. Extensive experiments show that MPGAN has significantly greater accuracy than state-of-the-art methods.

cs.CV↗

arXiv2020

Cascade Network with Guided Loss and Hybrid Attention for Two-view Geometry

In this paper, we are committed to designing a high-performance network for two-view geometry. We first propose a Guided Loss and theoretically establish the direct negative correlation between the loss and Fn-measure by dynamically adjusting the weights of positive and negative classes during training, so that the network is always trained towards the direction of increasing Fn-measure. By this way, the network can maintain the advantage of the cross-entropy loss while maximizing the Fn-measure. We then propose a hybrid attention block to extract feature, which integrates the bayesian attentive context normalization (BACN) and channel-wise attention (CA). BACN can mine the prior information to better exploit global context and CA can capture complex channel context to enhance the channel awareness of the network. Finally, based on our Guided Loss and hybrid attention block, a cascade network is designed to gradually optimize the result for more superior performance. Experiments have shown that our network achieves the state-of-the-art performance on benchmark datasets.

cs.CV↗

arXiv2020

Semi-Supervised Text Simplification with Back-Translation and Asymmetric Denoising Autoencoders

Text simplification (TS) rephrases long sentences into simplified variants while preserving inherent semantics. Traditional sequence-to-sequence models heavily rely on the quantity and quality of parallel sentences, which limits their applicability in different languages and domains. This work investigates how to leverage large amounts of unpaired corpora in TS task. We adopt the back-translation architecture in unsupervised machine translation (NMT), including denoising autoencoders for language modeling and automatic generation of parallel data by iterative back-translation. However, it is non-trivial to generate appropriate complex-simple pair if we directly treat the set of simple and complex corpora as two different languages, since the two types of sentences are quite similar and it is hard for the model to capture the characteristics in different types of sentences. To tackle this problem, we propose asymmetric denoising methods for sentences with separate complexity. When modeling simple and complex sentences with autoencoders, we introduce different types of noise into the training process. Such a method can significantly improve the simplification performance. Our model can be trained in both unsupervised and semi-supervised manner. Automatic and human evaluations show that our unsupervised model outperforms the previous systems, and with limited supervision, our model can perform competitively with multiple state-of-the-art simplification systems.

cs.CL↗

arXiv2020

Time-aware Gradient Attack on Dynamic Network Link Prediction

In network link prediction, it is possible to hide a target link from being predicted with a small perturbation on network structure. This observation may be exploited in many real world scenarios, for example, to preserve privacy, or to exploit financial security. There have been many recent studies to generate adversarial examples to mislead deep learning models on graph data. However, none of the previous work has considered the dynamic nature of real-world systems. In this work, we present the first study of adversarial attack on dynamic network link prediction (DNLP). The proposed attack method, namely time-aware gradient attack (TGA), utilizes the gradient information generated by deep dynamic network embedding (DDNE) across different snapshots to rewire a few links, so as to make DDNE fail to predict target links. We implement TGA in two ways: one is based on traversal search, namely TGA-Tra; and the other is simplified with greedy search for efficiency, namely TGA-Gre. We conduct comprehensive experiments which show the outstanding performance of TGA in attacking DNLP algorithms.

cs.SI↗

arXiv2020

Beamforming Optimization for Intelligent Reflecting Surface Assisted MIMO: A Sum-Path-Gain Maximization Approach

Recently, intelligent reflecting surface (IRS) has emerged as an appealing technique that enables wireless communications with low hardware cost and low power consumption. In this letter, we consider an IRS-assisted point-to-point multi-input multi-output (MIMO) system, where a source communicates with its destination with the help of an IRS. Our goal is to maximize the spectral efficiency of this system by jointly optimizing the (active) precoding at the source and the (passive) phase shifters (PSs) at the IRS. However, this turns out to be an intractable mixed integer non-convex optimization problem. To circumvent the intractability, we propose a new sum-path-gain maximization (SPGM) criterion to obtain a high-quality and efficient suboptimal solution to this problem. Specifically, the PSs are first designed based on a simplified optimization problem, which aims to maximize the sum-gains of the spatial paths between the source and the destination. Then, a low-complexity alternating direction method of multipliers (ADMM) algorithm is utilized to solve this simplified problem. Finally, with the above obtained PSs, the source precoding is derived by performing the singular value decomposition (SVD) on the effective channel between the source and the destination. Numerical results demonstrate that the proposed scheme can achieve near-optimal performance.

cs.IT↗

arXiv2020

Channel Estimation and Transmission for Intelligent Reflecting Surface Assisted THz Communications

Intelligent reflecting surface (IRS) is envisioned as a promising technology to broaden signal coverage and enhance transmission in terahertz (THz) communications. Due to the passivity of IRS, the channel measurement can not be achieved by traditional pilot manner and the subsequent cooperative transmission design remains an open problem. This paper investigates the channel estimation and transmission solutions for massive multiple input multiple output (MIMO) IRS-assisted THz system. The channel estimation is realized by beam training and the quantization error is analyzed for evaluating performance. In addition, a novel hierarchical search codebook design is proposed as a low-complexity basis of beam training. Based on above foundations, we propose a cooperative channel estimation procedure to tactfully acquire the channel knowledge. Finally, by leveraging obtained channel information, the designs of IRS and transceivers are directly provided in closed form without reconstructing the full channel matrix or additional optimization. Simulation and numerical results are presented to illustrate the minimum signal to noise ratio (SNR) required for beam training and the efficacy of the proposed transmission solutions.

cs.IT↗

arXiv2020

Conceptual Design and Preliminary Results of a VR-based Radiation Safety Training System for Interventional Radiologists

Recent studies have reported an increased risk of developing brain and neck tumors, as well as cataracts, in practitioners in interventional radiology (IR). Occupational radiation protection in IR has been a top concern for regulatory agencies and professional societies. To help minimize occupational radiation exposure in IR, we conceptualized a virtual reality (VR) based radiation safety training system to help operators understand complex radiation fields and to avoid high radiation areas through game-like interactive simulations. The preliminary development of the system has yielded results suggesting that the training system can calculate and report the radiation exposure after each training session based on a database precalculated from computational phantoms and Monte Carlo simulations and the position information provided in real-time by the MS Hololens headset worn by trainee. In addition, real-time dose rate and cumulative dose will be displayed to the trainee by MS Hololens to help them adjust their practice. This paper presents the conceptual design of the overall hardware and software design, as well as preliminary results to combine MS HoloLens headset and complex 3D X-ray field spatial distribution data to create a mixed reality environment for safety training purpose in IR.

physics.med-ph↗

arXiv2019

Terahertz Communications (TeraCom): Challenges and Impact on 6G Wireless Systems

Terahertz communications are envisioned as a key technology for 6G, which requires 100+ Gbps data rates, 1-millisecond latency, among other performance metrics. As a fundamental wireless infrastructure, the THz communication can boost abundant promising applications, including next-generation WLAN systems like Tera-WiFi, THz wireless backhaul, as well as other long-awaited novel communication paradigms. Serving as a basis of efficient wireless communication and networking design, this paper highlights the key THz channel features and recent advancements in device technologies. In light of these, impact and guidelines on 6G wireless communication and networking are elaborated. We believe the progress of THz technologies is helping finally close the so called THz Gap, and will realize THz communications as a pillar of 6G wireless systems.

eess.SP↗

arXiv2019

Reconfigurable Intelligent Surface-Based Wireless Communication: Antenna Design, Prototyping and Experimental Results

One of the key enablers of future wireless communications is constituted by massive multiple-input multiple-output (MIMO) systems, which can improve the spectral efficiency by orders of magnitude. However, in existing massive MIMO systems, conventional phased arrays are used for beamforming, which result in excessive power consumption and hardware cost. Recently, reconfigurable intelligent surface (RIS) has been considered as one of the revolutionary technologies to enable energy-efficient and smart wireless communications, which is a two-dimensional structure with a large number of passive elements. In this paper, we propose and develop a new type of high-gain yet low-cost RIS having 256 elements. The proposed RIS combines the functions of phase shift and radiation together on an electromagnetic surface, where positive intrinsic-negative (PIN) diodes are used to realize 2-bit phase shifting for beamforming. Based on this radical design, the world's first wireless communication prototype using RIS having 256 2-bit elements is designed and developed. Specifically, the prototype conceived consists of modular hardware and flexible software, including the hosts for parameter setting and data exchange, the universal software radio peripherals (USRPs) for baseband and radio frequency (RF) signal processing, as well as the RIS for signal transmission and reception. Our performance evaluation confirms the feasibility and efficiency of RISs in future wireless communications. More particularly, it is shown that a 21.7 dBi antenna gain can be obtained by the proposed RIS at 2.3 GHz, while at the millimeter wave (mmWave) frequency, i.e., 28.5 GHz, a 19.1 dBi antenna gain can be achieved. Furthermore, the over-the-air (OTA) test results show that the RIS-based wireless communication prototype developed is capable of significantly reducing the power consumption.

cs.IT↗

arXiv2019

Automatic phase space generation for Monte Carlo calculations of Intensity Modulated Particle Therapy

Monte Carlo (MC) is generally considered as the most accurate dose calculation tool for particle therapy. However, a proper description of the beam particles kinematics is a necessary input for a realistic simulation. Such a description can be stored in phase space (PS) files for different beam energies. A PS file contains kinetic information such as energies, positions and travelling directions for particles traversing a plane perpendicular to the beam direction. The accuracy of PS files play a critical role in the performance of the MC method for dose calculations. A PS file can be generated with a set of parameters describing analytically the beam kinematics. However, determining such parameters can be tedious and time consuming. Thus, we have developed an algorithm to obtain those parameters automatically and efficiently. In this paper, we present such an algorithm and compare dose calculations using PS automatically generated for the Shanghai Proton and Heavy Ion Center (SPHIC) with measurements. The gamma-index for comparing calculated depth dose distributions (DDD) with measurements are above 96.0% with criterion 0.6%/0.6 mm. For each single energy, the mean difference percentage between calculated lateral spot sizes at 5 different depths and measurements are below 3.5%.

physics.med-ph↗

arXiv2019

Perceptual-assisted Adversarial Adaptation for Choroid Segmentation in optical coherence tomography

Accurate choroid segmentation in optical coherence tomography (OCT) image is vital because the choroid thickness is a major quantitative biomarker of many ocular diseases. Deep learning has shown its superiority in the segmentation of the choroid region but subjects to the performance degeneration caused by the domain discrepancies (e.g., noise level and distribution) among datasets obtained from the OCT devices of different manufacturers. In this paper, we present an unsupervised perceptual-assisted adversarial adaptation (PAAA) framework for efficiently segmenting the choroid area by narrowing the domain discrepancies between different domains. The adversarial adaptation module in the proposed framework encourages the prediction structure information of the target domain to be similar to that of the source domain. Besides, a perceptual loss is employed for matching their shape information (the curvatures of Bruch's membrane and choroid-sclera interface) which can result in a fine boundary prediction. The results of quantitative experiments show that the proposed PAAA segmentation framework outperforms other state-of-the-art methods.

eess.IV↗