Search

Designing a query-efficient model extraction strategy to steal models from cloud-based platforms with black-box constraints remains a challenge, especially for language models. In a more realistic setting, a lack of information about the target model’s internal parameters, gradients, training data, or even confidence scores prevents attackers from easily copying the target model. Selecting informative and useful examples to train a substitute model is critical to query-efficient model stealing. We propose a novel model extraction framework that fine-tunes a pretrained model based on bidirectional encoder representations from transformers (BERT) while improving query efficiency by utilizing an active learning selection strategy. The active learning strategy, incorporating semantic-based diversity sampling and class-balanced uncertainty sampling, builds an informative subset from the public unannotated dataset as the input for fine-tuning. We apply our method to extract deep classifiers with identical and mismatched architectures as the substitute model under tight and moderate query budgets. Furthermore, we evaluate the transferability of adversarial examples constructed with the help of the models extracted by our method. The results show that our method achieves higher accuracy with fewer queries than existing baselines and the resulting models exhibit a high transferability success rate of adversarial examples.

Deep neural networks that play a pivotal role in fields such as images, text, and audio are vulnerable to adversarial attacks. In current textual adversarial attacks, the vast majority are configured with a black-box soft-label which is achieved by the gradient information or confidence of the model. Therefore, it becomes challenging and realistic to implement adversarial attacks using only the predicted top labels of the hard-label model. Existing methods to implement hard-label adversarial attacks use population-based genetic optimization algorithms. However, this approach requires significant query consumption, which is a considerable shortcoming. To solve this problem, we propose a new textual black-box hard-label adversarial attack algorithm based on the idea of differential evolution of populations, called the text-based differential evolution (TDE) algorithm. First, the method will judge the importance of the words of the initial rough adversarial examples, according to which only the keywords in the text sentence will be operated, and the rest of the words will be gradually replaced with the original words so as to reduce the words in the sentence in which the replacement occurs. Our method judges the quality of semantic similarity of the adversarial examples in the replacement process and deposits high-quality adversarial example individuals into the population. Secondly, the optimization process of adversarial examples is combined and optimized according to the word importance. Compared with existing methods based on genetic algorithm guidance, our method avoids a large number of meaningless repetitive queries and significantly improves the overall attack efficiency of the algorithm and the semantic quality of the generated adversarial examples. We experimented with multiple datasets on three text tasks of sentiment classification, natural language inference, and toxic comment, and also perform experimental comparisons on models and APIs in realistic scenarios. For example, in the Google Cloud commercial API adversarial attack experiment, compared to the existing hard-label method, our method reduces the average number of queries required for the attack from 6986 to 176, and increases semantic similarity from 0.844 to 0.876. It is shown through extensive experimental data that our approach not only significantly reduces the number of queries, but also significantly outperforms existing methods in terms of the quality of adversarial examples.

Abstract The low viability during gastrointestinal transit and poor mucoadhesion considerably limits the effectiveness of Ligilactobacillus salivarius Li01 (Li01) in regulating gut microbiota and alleviating inflammatory bowel disease (IBD). In this study, a delivery system was designed through layer-by-layer (LbL) encapsulating a single Li01cell with chitosan and alginate. The layers were strengthened by cross-linking to form a firm and mucoadhesive shell (~10 nm thickness) covering the bacterial cell. The LbL Li01 displayed improved viability under simulated gastrointestinal conditions and mucoadhesive function. Almost no cells could be detected among the free Li01 after 2 h incubation in digestive fluids, while for LbL Li01, the total reduction was around 3 log CFU/mL and the viable number of cells remained above 6 log CFU/mL. Besides, a 5-fold increase in the value of rupture length and a two-fold increase in the number of peaks were found in the (bacteria-mucin) adhesion curves of LbL Li01, compared to those of free Li01. Oral administration with LbL Li01 on colitis mice facilitated intestinal barrier recovery and restoration of the gut microbiota. The improved functionality of Li01 by LbL encapsulation could increase the potential for the probiotic to be used in clinical applications to treat IBD; this should be explored in future studies.

Adversarial examples (AEs) against deep neural networks (DNNs) raise wide concerns about the robustness of DNNs. Existing detection mechanisms are often limited to a given attack algorithm. Therefore, it is highly desirable to develop a robust detection approach that remains effective for a large group of attack algorithms. In addition, most of the existing defences only perform well for small images (e.g. MNIST and Canadian institute for advanced research (CIFAR)) rather than large images (e.g. ImageNet). In this paper, the authors propose a robust and effective defence method for analysing the sensitivity of various AEs, especially in a much harder case (large images). Their method first creates a feature map from the input space to the new feature space, by utilising 19 different feature mapping methods. Then, a detector is learned with the machine‐learning algorithm to recognise the unique distribution of AEs. Their extensive evaluations on their proposed detector show that their detector can achieve: (i) low false‐positive rate (

Online short-term rental platforms, such as Airbnb, have been becoming popular, and a better pricing strategy is imperative for hosts of new listings. In this paper, we analyzed the relationship between the description of each listing and its price, and proposed a text-based price recommendation system called TAPE to recommend a reasonable price for newly added listings. We used deep learning techniques (e.g., feedforward network, long short-term memory, and mean shift) to design and implement TAPE. Using two chronologically extracted datasets of the same four cities, we revealed important factors (e.g., indoor equipment and high-density area) that positively or negatively affect each property’s price, and evaluated our preliminary and enhanced models. Our models achieved a Root-Mean-Square Error (RMSE) of 33.73 in Boston, 20.50 in London, 34.68 in Los Angeles, and 26.31 in New York City, which are comparable to an existing model that uses more features.

Social networks are important media for spreading information, ideas, and influence among individuals. Most existing research focuses on understanding the characteristics of social networks, investigating how information is spread through the "word-of-mouth" effect of social networks, or exploring social influences among individuals and groups. However, most studies ignore negative influences among individuals and groups. Motivated by the goal of alleviating social problems, such as drinking, smoking, and gambling, and influence-spreading problems, such as promoting new products, we consider positive and negative influences, and propose a new optimization problem called the Minimum-sized Positive Influential Node Set (MPINS) selection problem to identify the minimum set of influential nodes such that every node in the network can be positively influenced by these selected nodes with no less than a threshold of θ. Our contributions are threefold. First, we prove that, under the independent cascade model considering positive and negative influences, MPINS is APX-hard. Subsequently, we present a greedy approximation algorithm to address the MPINS selection problem. Finally, to validate the proposed greedy algorithm, we conduct extensive simulations and experiments on random graphs and seven different real-world data sets that represent small-, medium-, and large-scale networks.

Image captchas have recently become very popular and are widely deployed across the Internet to defend against abusive programs. However, the ever-advancing capabilities of computer vision have gradually diminished the security of image captchas and made them vulnerable to attack. In this paper, we first classify the currently popular image captchas into three categories: selection-based captchas, slide-based captchas, and click-based captchas. Second, we propose simple yet powerful attack frameworks against each of these categories of image captchas. Third, we systematically evaluate our attack frameworks against 10 popular real-world image captchas, including captchas from tencent.com, google.com, and 12306.cn. Fourth, we compare our attacks against nine online image recognition services and against human labors from eight underground captcha-solving services. Our evaluation results show that (1) each of the popular image captchas that we study is vulnerable to our attacks; (2) our attacks yield the highest captcha-breaking success rate compared with state-of-the-art methods in almost all scenarios; and (3) our attacks achieve almost as high a success rate as human labor while being much faster. Based on our evaluation, we identify some design flaws in these popular schemes, along with some best practices and design principles for more secure captchas. We also examine the underground market for captcha-solving services, identifying 152 such services. We then seek to measure this underground market with data from these services. Our findings shed light on understanding the scale, impact, and commercial landscape of the underground market for captcha solving.

Despite their tremendous success in various machine learning tasks, deep neural networks (DNNs) are inherently vulnerable to adversarial examples, which are maliciously crafted inputs to cause DNNs to misbehave. Intensive research has been conducted on this phenomenon in simple tasks (e.g., image classification). However, little is known about this adversarial vulnerability for object detection, a much more complicated task, which often requires specialized DNNs and multiple additional components. In this paper, we present DETECTSEC, a uniform platform for robustness analysis of object detection models. Currently, DETECTSEC implements 13 representative adversarial attacks with 7 utility metrics and 13 defenses on 18 standard object detection models. Leveraging DETECTSEC, we conduct the first rigorous evaluation of adversarial attacks on the state-of-the-art object detection models. We analyze the impact of the factors including DNN architecture and capacity on the model robustness. We show that many conclusions about adversarial attacks and defenses in image classification tasks do not transfer to object detection tasks, for example, the targeted attack is stronger than the untargeted attack for two-stage detectors. Our findings will aid future efforts in understanding and defending against adversarial attacks in complicated tasks. In addition, we compare the robustness of different detection models and discuss their relative strengths and weaknesses. The platform DETECTSEC will be open source as a unique facility for further research on adversarial attacks and defenses in object detection tasks. [ABSTRACT FROM AUTHOR]

Due to the coupling of multiple nonlinear factors in the position control system of the double closed-loop digital hydraulic cylinder, achieving precise and stable control of complex nonlinear systems with a single structural controller is difficult. To address this problem, we propose a Global Fast Terminal Sliding Mode Active Disturbance Rejection Control (GFTSMADRC). Firstly, by focusing on the mathematical model of the current double closed-loop digital hydraulic cylinder, the high-order nonlinear state equation of the system is derived, and employing active disturbance rejection control (ADRC), the high-order nonlinear system is transformed into a second-order integral series control system. Secondly, we propose Sliding Mode Active Disturbance Rejection Control (SMADRC) by integrating sliding mode variable structure control. Thirdly, to further enhance system performance, we propose a GFTSMADRC strategy by incorporating global fast terminal sliding mode variable structure control. The control law and adaptive law of the system are derived based on the Lyapunov stability theorem. Specifically, the effectiveness of the proposed control method is validated through simulation and experimentation.

Crowdsourcing has attracted much attention for its convenience to collect labels from non-expert workers instead of experts. However, due to the high level of noise fromthe non-experts, a label aggregationmodel that infers the true label from noisy crowdsourced labels is required. In this article, we propose a novel framework based on graph neural networks for aggregating crowd labels.We construct a heterogeneous graph between workers and tasks and derive a new graph neural network to learn the representations of nodes and the true labels. Besides, we exploit the unknown latent interaction between the same type of nodes (workers or tasks) by adding a homogeneous attention layer in the graph neural networks. Experimental results on 13 real-world datasets show superior performance over state-of-the-art models. [ABSTRACT FROM AUTHOR]

Code retrieval is to find the code snippet from a large corpus of source code repositories that highly matches the query of natural language description. Recent work mainly uses natural language processing techniques to process both query texts (i.e., human natural language) and code snippets (i.e., machine programming language), however, neglecting the deep structured features of query texts and source codes, both of which contain rich semantic information. In this article, we propose an end-to-end deep graph matching and searching (DGMS) model based on graph neural networks for the task of semantic code retrieval. To this end, we first represent both natural language query texts and programming language code snippets with the unified graphstructured data, and then use the proposed graphmatching and searching model to retrieve the best matching code snippet. In particular, DGMS not only captures more structural information for individual query texts or code snippets, but also learns the fine-grained similarity between them by cross-attention based semantic matching operations. We evaluate the proposed DGMS model on two public code retrieval datasets with two representative programming languages (i.e., Java and Python). Experiment results demonstrate that DGMS significantly outperforms state-of-the-art baseline models by a large margin on both datasets. Moreover, our extensive ablation studies systematically investigate and illustrate the impact of each part of DGMS. [ABSTRACT FROM AUTHOR]

In this paper, we redesign, implement, and evaluate ShareSafe (Based on SecGraph), an open-source secure graph data sharing/publishing platform. Within ShareSafe, we propose De-anonymization Quantification Module and Recommendation Module. Besides, we model the attackers' background knowledge and evaluate the relation between graph data privacy and the structure of the graph. To the best of our knowledge, ShareSafe is the first platform that enables users to perform data perturbation, utility evaluation, De-A evaluation, and Privacy Quantification. Leveraging ShareSafe, we conduct a more comprehensive and advanced utility and privacy evaluation. The results demonstrate that (1) The risk of privacy leakage of anonymized graph increases with the attackers' background knowledge. (2) For a successful de-anonymization attack, the seed mapping, even relatively small, plays a much more important role than the auxiliary graph. (3) The structure of graph has a fundamental and significant effect on the utility and privacy of the graph. (4) There is no optimal anonymization/de-anonymization algorithm. For different environment, the performance of each algorithm varies from each other. [ABSTRACT FROM AUTHOR]

The nonlinear factors in the digital hydraulic cylinder will reduce the accuracy of the control system. In order to improve the control accuracy of the control system, in this paper, a model reference adaptive disturbance rejection control method based on neural network is proposed. Firstly, the dead zone model is used to replace the nonlinear link in the feedback mechanism. A detailed mathematical model of digital hydraulic cylinder is established and the nonlinear hydraulic spring force is also considered, and a complete nonlinear state space model of digital hydraulic cylinder is derived based on LuGre friction model. Then the reference model is designed. By introducing ESO (extended state observer), the uncertainties and external disturbances of the controlled object are all equivalent to a total disturbance. The RBF (Radial Basis Function) network is used to approximate the unknown function FZ, the neural model reference adaptive disturbance rejection composite controller is designed by using Lyapunov direct method and Barbalat lemma. Finally, the simulation verification is carried out by using MATLAB. The simulation results show that the control strategy can effectively improve the response characteristics of the system, reduce the steady-state error of the system, and improve the robustness of the system.

When people utilize social applications and services, their privacy suffers a potential serious threat. In this article, we present a novel, robust, and effective de-anonymization attack to mobility trace data and social data. First, we design a Unified Similarity (US) measurement, which takes account of local and global structural characteristics of data, information obtained from auxiliary data, and knowledge inherited from ongoing de-anonymization results. By analyzing the measurement on real datasets, we find that some data can potentially be de-anonymized accurately and the other can be de-anonymized in a coarse granularity. Utilizing this property, we present a US-based De-Anonymization (DA) framework, which iteratively deanonymizes data with accuracy guarantee. Then, to de-anonymize large-scale data without knowledge of the overlap size between the anonymized data and the auxiliary data, we generalize DA to an Adaptive De-Anonymization (ADA) framework. By smartly working on two core matching subgraphs, ADA achieves high de-anonymization accuracy and reduces computational overhead. Finally, we examine the presented de-anonymization attack on three well-known mobility traces: St Andrews, Infocom06, and Smallblue, and three social datasets: ArnetMiner, Google+, and Facebook. The experimental results demonstrate that the presented de-anonymization framework is very effective and robust to noise. [ABSTRACT FROM AUTHOR]

This paper aims to eliminate nonlinear friction from the performance of the digital hydraulic cylinder to enable it to have good adaptive ability. First, a mathematical model of a digital hydraulic cylinder based on the LuGre friction model was established, and then a dual-observer structure was designed to estimate the unobservable state variables in the friction model. The Lyapunov method is used to prove the global asymptotic stability of the closed-loop system using the adaptive friction compensation method. Finally, Simulink is used to simulate the system performance. The simulation results indicate that the addition of adaptive friction compensation control can effectively reduce system static error, suppress system limit loop oscillation, “position decapitation,” “speed dead zone,” and low-speed creep phenomena, and improve the overall performance of the digital hydraulic cylinder. The control method has practical application value for improving the performance index of the digital hydraulic cylinder.

Minimum Connected Dominating Sets (MCDSs) are used as virtual backbones for efficient routing and broadcasting in wireless networks extensively. However, the MCDS problem is NP-Complete even in Unit Disk Graphs. Therefore, many heuristic-based approximation algorithms have been proposed recently. In these approaches, networks are deterministic where two nodes are assumed either connected or disconnected. In most real applications, however, there are many intermittently connected wireless links called lossy links, which only provide probabilistic connectivity. For wireless networks with lossy links, we propose a Probabilistic Network Model (PNM). Under this model, we measure the quality of Connected Dominating Sets (CDSs) using CDS reliability defined as the minimum upper limit of the node-to-node delivery ratio between any pair of dominators in a CDS. We attempt to construct a MCDS while its reliability is above a preset application-specified threshold σ, called σ-Reliable MCDS (σ-RMCDS). We claim that constructing a σ-RMCDS is NP-Hard under the PNM model. We propose a novel Genetic Algorithm (GA) with immigrants schemes called RMCDS-GA to solve the σ-RMCDS problem. To evaluate the performance of RMCDS-GA, we conduct comprehensive simulations. The simulation results show that compared with the traditional MCDS algorithms. RMCDS-GA can construct a more reliable CDS without increasing the size of a CDS. [ABSTRACT FROM AUTHOR]

Data collection is a common operation of wireless sensor networks (WSNs). The performance of data collection can be measured by its achievable network capacity. However, most existing works focus on the network capacity of unicast, multicast or/and broadcast. In this article, we study the snapshot/continuous data collection (SDC/CDC) problem under the physical interference model for randomly deployed dense WSNs. For SDC, we propose a Cell-Based Path Scheduling (CBPS) algorithm based on network partitioning. Theoretical analysis shows that its achievable network capacity is order-optimal. For CDC, a novel Segment- Based Pipeline Scheduling (SBPS) algorithm is proposed which combines the pipeline technique and the compressive data gathering technique. Theoretical analysis shows that SBPS significantly speeds up the CDC process and achieves a high network capacity. [ABSTRACT FROM AUTHOR]