arXiv subjects

Ru Zhang

Publications and source records attributed to Ru Zhang.

At least 55 records · Page 3Linked to original sources

arXiv2024

An Enhanced Prompt-Based LLM Reasoning Scheme via Knowledge Graph-Integrated Collaboration

While Large Language Models (LLMs) demonstrate exceptional performance in a multitude of Natural Language Processing (NLP) tasks, they encounter challenges in practical applications, including issues with hallucinations, inadequate knowledge updating, and limited transparency in the reasoning process. To overcome these limitations, this study innovatively proposes a collaborative training-free reasoning scheme involving tight cooperation between Knowledge Graph (KG) and LLMs. This scheme first involves using LLMs to iteratively explore KG, selectively retrieving a task-relevant knowledge subgraph to support reasoning. The LLMs are then guided to further combine inherent implicit knowledge to reason on the subgraph while explicitly elucidating the reasoning process. Through such a cooperative approach, our scheme achieves more reliable knowledge-based reasoning and facilitates the tracing of the reasoning results. Experimental results show that our scheme significantly progressed across multiple datasets, notably achieving over a 10% improvement on the QALD10 dataset compared to the best baseline and the fine-tuned state-of-the-art (SOTA) work. Building on this success, this study hopes to offer a valuable reference for future research in the fusion of KG and LLMs, thereby enhancing LLMs' proficiency in solving complex issues.

cs.CL↗

arXiv2024

UP4LS: User Profile Constructed by Multiple Attributes for Enhancing Linguistic Steganalysis

Linguistic steganalysis (LS) tasks aim to detect whether a text contains secret information. Existing LS methods focus on the deep-learning model design and they achieve excellent results in ideal data. However, they overlook the unique user characteristics, leading to weak performance in social networks. And a few stegos here that further complicate detection. We propose the UP4LS, a framework with the User Profile for enhancing LS in realistic scenarios. Three kinds of user attributes like writing habits are explored to build the profile. For each attribute, the specific feature extraction module is designed. The extracted features are mapped to high-dimensional user features via the deep-learning model of the method to be improved. The content feature is extracted by the language model. Then user and content features are integrated. Existing methods can improve LS results by adding the UP4LS framework without changing their deep-learning models. Experiments show that UP4LS can significantly enhance the performance of LS-task baselines in realistic scenarios, with the overall Acc increased by 25%, F1 increased by 51%, and SOTA results. The improvement is especially pronounced in fewer stegos. Additionally, UP4LS also sets the stage for the related-task SOTA methods to efficient LS.

cs.CL↗

arXiv2024

Pseudo-label Based Domain Adaptation for Zero-Shot Text Steganalysis

Currently, most methods for text steganalysis are based on deep neural networks (DNNs). However, in real-life scenarios, obtaining a sufficient amount of labeled stego-text for correctly training networks using a large number of parameters is often challenging and costly. Additionally, due to a phenomenon known as dataset bias or domain shift, recognition models trained on a large dataset exhibit poor generalization performance on novel datasets and tasks. Therefore, to address the issues of missing labeled data and inadequate model generalization in text steganalysis, this paper proposes a cross-domain stego-text analysis method (PDTS) based on pseudo-labeling and domain adaptation (unsupervised learning). Specifically, we propose a model architecture combining pre-trained BERT with a single-layer Bi-LSTM to learn and extract generic features across tasks and generate task-specific representations. Considering the differential contributions of different features to steganalysis, we further design a feature filtering mechanism to achieve selective feature propagation, thereby enhancing classification performance. We train the model using labeled source domain data and adapt it to target domain data distribution using pseudo-labels for unlabeled target domain data through self-training. In the label estimation step, instead of using a static sampling strategy, we propose a progressive sampling strategy to gradually increase the number of selected pseudo-label candidates. Experimental results demonstrate that our method performs well in zero-shot text steganalysis tasks, achieving high detection accuracy even in the absence of labeled data in the target domain, and outperforms current zero-shot text steganalysis methods.

cs.CV↗

arXiv2022

LogKernel A Threat Hunting Approach Based on Behaviour Provenance Graph and Graph Kernel Clustering

Cyber threat hunting is a proactive search process for hidden threats in the organization's information system. It is a crucial component of active defense against advanced persistent threats (APTs). However, most of the current threat hunting methods rely on Cyber Threat Intelligence(CTI), which can find known attacks but cannot find unknown attacks that have not been disclosed by CTI. In this paper, we propose LogKernel, a threat hunting method based on graph kernel clustering which can effectively separates attack behaviour from benign activities. LogKernel first abstracts system audit logs into Behaviour Provenance Graphs (BPGs), and then clusters graphs by embedding them into a continuous space using a graph kernel. In particular, we design a new graph kernel clustering method based on the characteristics of BPGs, which can capture structure information and rich label information of the BPGs. To reduce false positives, LogKernel further quantifies the threat of abnormal behaviour. We evaluate LogKernel on the malicious dataset which includes seven simulated attack scenarios and the DAPRA CADETS dataset which includes four attack scenarios. The result shows that LogKernel can hunt all attack scenarios among them, and compared to the state-of-the-art methods, it can find unknown attacks.

cs.CR↗

arXiv2021

On-chip erbium-doped lithium niobate microring lasers

Lithium niobate on insulator (LNOI), regarded as an important candidate platform for optical integration due to its excellent nonlinear, electro-optic and other physical properties, has become a research hotspot. Light source, as an essential component for integrated optical system, is urgently needed. In this paper, we reported the realization of 1550-nm band on-chip LNOI microlasers based on erbium-doped LNOI ring cavities with loaded quality factors higher than one million, which were fabricated by using electron beam lithography and inductively coupled plasma reactive ion etching processes. These microlasers demonstrated a low pump threshold of ~20 μW and stable performance under the pump of a 980-nm band continuous laser. Comb-like laser spectra spanning from 1510 nm to 1580 nm were observed in high pump power regime, which lays the foundation of the realization of pulsed laser and frequency combs on rare-earth ion doped LNOI platform. This work has effectively promoted the development of on-chip integrated active LNOI devices.

physics.optics↗

arXiv2021

Integrated LNOI single-mode lasers by Vernier effect

Microcavity lasers based on erbium-doped lithium niobate on insulator (LNOI), which are key devices for LNOI integrated photonics, have attracted much attention recently. In this Letter, we report the realization of a C-band single-mode laser using Vernier effect in two coupled Erbium-doped LNOI microrings with different radii under the pump of a 980-nm continuous laser. The laser, operating stably over a large range of pumping power, has a pump threshold of ~200 μW and a side-mode suppression ratio exceeding 26 dB. The high-performance LNOI single-mode laser will promote the development of lithium niobate integrated photonics.

physics.optics↗

arXiv2021

On-chip erbium-doped lithium niobate waveguide amplifiers

Lithium niobate on insulator (LNOI), as an emerging and promising optical integration platform, faces shortages of on-chip active devices including lasers and amplifiers. Here, we report the fabrication on-chip erbium-doped LNOI waveguide amplifiers based on electron beam lithography and inductively coupled plasma reactive ion etching. A net internal gain of ~30 dB/cm in communication band was achieved in the fabricated waveguide amplifiers under the pump of a 974-nm continuous laser. This work develops new active devices on LNOI and will promote the development of LNOI integrated photonics.

physics.optics↗

arXiv2020

Construction of Two Statistical Anomaly Features for Small-Sample APT Attack Traffic Classification

Advanced Persistent Threat (APT) attack, also known as directed threat attack, refers to the continuous and effective attack activities carried out by an organization on a specific object. They are covert, persistent and targeted, which are difficult to capture by traditional intrusion detection system(IDS). The traffic generated by the APT organization, which is the organization that launch the APT attack, has a high similarity, especially in the Command and Control(C2) stage. The addition of features for APT organizations can effectively improve the accuracy of traffic detection for APT attacks. This paper analyzes the DNS and TCP traffic of the APT attack, and constructs two new features, C2Load_fluct (response packet load fluctuation) and Bad_rate (bad packet rate). The analysis showed APT attacks have obvious statistical laws in these two features. This article combines two new features with common features to classify APT attack traffic. Aiming at the problem of data loss and boundary samples, we improve the Adaptive Synthetic(ADASYN) Sampling Approach and propose the PADASYN algorithm to achieve data balance. A traffic classification scheme is designed based on the AdaBoost algorithm. Experiments show that the classification accuracy of APT attack traffic is improved after adding new features to the two datasets so that 10 DNS features, 11 TCP and HTTP/HTTPS features are used to construct a Features set. On the two datasets, F1-score can reach above 0.98 and 0.94 respectively, which proves that the two new features in this paper are effective for APT traffic detection.

cs.CR↗

arXiv2020

Constructing feature variation coefficients to evaluate feature learning capabilities of convolutional layers in steganographic detection algorithms of spatial domain

Traditional steganalysis methods generally include two steps: feature extraction and classification.A variety of steganalysis algorithms based on CNN (Convolutional Neural Network) have appeared in recent years. Among them, the convolutional layer of the CNN model is usually used to extract steganographic features, and the fully connected layer is used for classification. Because the effectiveness of feature extraction seriously influences the accuracy of classification, designers generally improve the accuracy of steganographic detection by improving the convolutional layer. For example, common optimizing methods in convolutional layer include the improvement of convolution kernel, activation functions, pooling functions, network structures, etc. However, due to the complexity and unexplainability of convolutional layers, it is difficult to quantitatively analyze and compare the effectiveness of feature extraction. Therefore, this paper proposes the variation coefficient to evaluate the feature learning ability of convolutional layers. We select four typical image steganalysis models based CNN in spatial domain, such as Ye-Net, Yedroudj-Net, Zhu-Net, and SR-Net as use cases, and verify the validity of the variation coefficient through experiments. Moreover, according to the variation coefficient , a features modification layer is used to optimize the features before the fully connected layer of the CNN model , and the experimental results show that the detection accuracy of the four algorithms were improved differently.

cs.CR↗

arXiv2020

On-chip erbium-doped lithium niobate microdisk lasers

Erbium-doped lithium niobate high-Q microdisk cavities were fabricated in batches by UV exposure, inductively coupled plasma reactive ion etching and chemo-mechanical polishing. The stimulated emission at 1531.6 nm was observed under the pump of a narrow-band laser working at 974 nm in erbium-doped lithium niobate microdisk cavity with threshold down to 400 μW and a conversion efficiency of 3.1{\times}10^{-4} %, laying the foundation for the LNOI integrated light source research.

physics.optics↗

arXiv2019

A Theoretical Study of Process Dependence for Critical Statistics in Standard Serial Models and Standard Parallel Models

Critical parts of the definitions of standard serial and standard parallel modes refer to stochastic independence. Standard serial models are defined by stochastic independence and identical distributions of their processing times. Processing times in the serial models are identical to the intercompletion time statistics. Similarly, standard parallel models assume stochastically independent and identical processing times. Their processing times are equivalent to the statistic known as total completion times. Little is known about what standard serial models can predict for the total completion time or what standard parallel models can predict for the intercompletion times. Here we demonstrate that standard serial models possess a tendency to predict a positive dependence for the total completion times with that always being true in the case of a single processing order. However, with mixtures of processing orders, standard serial models may predict negative dependence of the total completion times. Comparably, standard parallel models typically predict neither independence of the intercompletion times nor identical distributions. In fact, standard parallel models predict increasing intercompletion times as the individual channels continue to finish. Nevertheless, dramatically increasing hazard functions of the channels can defeat that tendency. And, standard parallel models can predict intercompletion time independence but only when individual channel distributions are exponential. Finally, we use these and ancillary mathematical results to conclude that standard serial and standard parallel models can never perfectly mimic one another. Therefore, our findings set the stage for explicit model testing between these classes.

q-bio.NC↗

arXiv2018

A Sequential Design Approach for Calibrating a Dynamic Population Growth Model

A comprehensive understanding of the population growth of a variety of pests is often crucial for efficient crop management. Our motivating application comes from calibrating a two-delay blowfly (TDB) model which is used to simulate the population growth of Panonychus ulmi (Koch) or European red mites that infest on apple leaves and diminish the yield. We focus on the inverse problem, that is, to estimate the set of parameters/inputs of the TDB model that produces the computer model output matching the field observation as closely as possible. The time series nature of both the field observation and the TDB outputs makes the inverse problem significantly more challenging than in the scalar valued simulator case. In spirit, we follow the popular sequential design framework of computer experiments. However, due to the time-series response, a singular value decomposition based Gaussian process model is used for the surrogate model, and subsequently, a new expected improvement criterion is developed for choosing the follow-up points. We also propose a new criterion for extracting the optimal inverse solution from the final surrogate. Three simulated examples and the real-life TDB calibration problem have been used to demonstrate higher accuracy of the proposed approach as compared to popular existing techniques.

stat.ME↗

arXiv2018

Invisible Steganography via Generative Adversarial Networks

Nowadays, there are plenty of works introducing convolutional neural networks (CNNs) to the steganalysis and exceeding conventional steganalysis algorithms. These works have shown the improving potential of deep learning in information hiding domain. There are also several works based on deep learning to do image steganography, but these works still have problems in capacity, invisibility and security. In this paper, we propose a novel CNN architecture named as \isgan to conceal a secret gray image into a color cover image on the sender side and exactly extract the secret image out on the receiver side. There are three contributions in our work: (i) we improve the invisibility by hiding the secret image only in the Y channel of the cover image; (ii) We introduce the generative adversarial networks to strengthen the security by minimizing the divergence between the empirical probability distributions of stego images and natural images. (iii) In order to associate with the human visual system better, we construct a mixed loss function which is more appropriate for steganography to generate more realistic stego images and reveal out more better secret images. Experiment results show that ISGAN can achieve start-of-art performances on LFW, Pascal VOC2012 and ImageNet datasets.

cs.MM↗

arXiv2018

Testing Selective Influence Directly Using Trackball Movement Tasks

Systems factorial technology (SFT; Townsend & Nozawa, 1995) is regarded as a useful tool to diagnose if features (or dimensions) of the investigated stimulus are processed in a parallel or serial fashion. In order to use SFT, one has to assume the speed to process each feature is influenced by that feature only, termed as selective influence (Sternberg, 1969). This assumption is usually untestable as the processing time for a stimulus feature is not observable. Stochastic dominance is traditionally used as an indirect evidence for selective influence (e.g., Townsend & Fifić, 2004). However, one should keep in mind that selective influence may be violated even when stochastic dominance holds. The current study proposes a trackball movement paradigm for a direct test of selective influence. The participants were shown a reference stimulus and a test stimulus simultaneously on a computer screen. They were asked to use the trackball to adjust the test stimulus until it appeared to match the position or shape of the reference stimulus. We recorded the reaction time, the parameters defined the reference stimulus (denoted as αand β), and the parameters defined the test stimulus (denoted as A and B). We tested selective influence of αand βon the amount of time to adjust A and B through testing selective influence of αand βon the values of A and B using the linear feasibility test (Dzhafarov & Kujala, 2010). We found that when the test was passed and stochastic dominance held, the inferred architecture was as expected, which was further confirmed by the trajectory of A and B observed in each trial. However, with stochastic dominance only SFT can suggest a prohibited architecture. Our results indicate the proposed method is more reliable for testing selective influence on the processing speed than examining stochastic dominance only.

stat.AP↗

arXiv2018

Efficient feature learning and multi-size image steganalysis based on CNN

For steganalysis, many studies showed that convolutional neural network has better performances than the two-part structure of traditional machine learning methods. However, there are still two problems to be resolved: cutting down signal to noise ratio of the steganalysis feature map and steganalyzing images of arbitrary size. Some algorithms required fixed size images as the input and had low accuracy due to the underutilization of the noise residuals obtained by various types of filters. In this paper, we focus on designing an improved network structure based on CNN to resolve the above problems. First, we use 3x3 kernels instead of the traditional 5x5 kernels and optimize convolution kernels in the preprocessing layer. The smaller convolution kernels are used to reduce the number of parameters and model the features in a small local region. Next, we use separable convolutions to utilize channel correlation of the residuals, compress the image content and increase the signal-to-noise ratio (between the stego signal and the image signal). Then, we use spatial pyramid pooling (SPP) to aggregate the local features, enhance the representation ability of features, and steganalyze arbitrary size image. Finally, data augmentation is adopted to further improve network performance. The experimental results show that the proposed CNN structure is significantly better than other four methods such as SRM, Ye-Net, Xu-Net, and Yedroudj-Net, when it is used to detect two spatial algorithms such as WOW and S-UNIWARAD with a wide variety of datasets and payloads.

cs.MM↗

arXiv2018

Local Gaussian Process Model for Large-scale Dynamic Computer Experiments

The recent accelerated growth in the computing power has generated popularization of experimentation with dynamic computer models in various physical and engineering applications. Despite the extensive statistical research in computer experiments, most of the focus had been on the theoretical and algorithmic innovations for the design and analysis of computer models with scalar responses. In this paper, we propose a computationally efficient statistical emulator for a large-scale dynamic computer simulator (i.e., simulator which gives time series outputs). The main idea is to first find a good local neighbourhood for every input location, and then emulate the simulator output via a singular value decomposition (SVD) based Gaussian process (GP) model. We develop a new design criterion for sequentially finding this local neighbourhood set of training points. Several test functions and a real-life application have been used to demonstrate the performance of the proposed approach over a naive method of choosing local neighbourhood set using the Euclidean distance among design points.

stat.ME↗

arXiv2017

A Theoretical Study of Process Dependence for Standard Two-Process Serial Models and Standard Two-Process Parallel Models

In this article we differentiate and characterize the standard two-process serial models and the standard two process parallel models by investigating the behavior of (conditional) distributions of the total completion times and survivals of intercompletion times without assuming any particular forms for the distributions of processing times. We address our argument through mathematical proofs and computational methods. It is found that for the standard two-process serial models, positive dependence between the total completion times does not hold if no specific distributional forms are imposed to the processing times. By contrast, for the standard two-process parallel models the total completion times are independent. According to different nature of process dependence, one can distinguish a standard two process serial model from a standard two-process parallel model. We also find that in standard two-process parallel models the monotonicity of survival function of the intercompletion time of stage 2 conditional on the completion of stage 1 depends on the monotonicity of the hazard function of processing time. We also find that the survival of intercompletion time(s) from stage 1 to stage 2 is increasing when the ratio of hazard function meets certain criterion. Then the empirical finding that the intercompletion time is grown with the growth of the number of recalled words can be accounted by standard parallel models. We also find that if the cumulative hazard function is concave or linear, the survival from stage 1 to stage 2 is increasing.

stat.AP↗

arXiv2017

Multiferroic Core-Shell Nanofibers, Assembly in a Magnetic field and Studies on MagnetoElectric Interactions

Ferromagnetic-ferroelectric nanocomposites are of interest for realizing strong strain mediated coupling between electric and magnetic subsystems due to high surface area-to-volume ratio. This report is on the synthesis of nickel ferrite (NFO) -barium titanate (BTO) core-shell nano-fibers, magnetic field assisted assembly into superstructures, and studies on magneto-electric (ME) interactions. Electrospinning techniques were used to prepare coaxial fibers of 0.5-1.5 micron in diameter. The core-shell structure of annealed fibers was confirmed by electron microscopy and scanning probe microscopy. The fibers were assembled into discs and films in a uniform magnetic field or a field gradient. Studies on ME coupling in the assembled films and discs were done by magnetic field H induced polarization, magneto-dielectric effects at low frequencies and at 16-24 GHz, and low frequency ME voltage coefficients (MEVC). We measured 2~ 2-7% change in remnant polarization and in the permittivity for H = 7 kOe, and a MEVC of 0.4 mV/cm Oe at 30 Hz. A model has been developed for low-frequency ME effects in an assembly of fibers and takes into account dipole-dipole interactions between the fibers and fiber discontinuity. Theoretical estimates for the low-frequency MEVC have been compared with the data. These results indicate strong ME coupling in superstructures of the core-shell fibers.

cond-mat.mtrl-sci↗