Search

Cyber-Physical Systems (CPS) is a security real-time embedded system. CPS integrates the information sensed by the current physical sensors, through high-speed real-time transmission, and then carries out powerful information processing to effectively interact and integrate the physical and the information worlds. With the aim to improve the quality of service, optimize the existing physical space, and increase security, collaborative filtering algorithms have also been widely used in various recommendation models for Internet of Things (IoT) services. However, general collaborative filtering algorithms cannot capture complex interactive information in the sparse Mashup-Web service call matrix, which leads to lower recommendation performance. Based on artificial intelligence technology, this study proposes a recommendation algorithm for security collaborative filtering service that integrates content similarity. A security collaborative filtering module is used to capture the complex interaction information between Mashup and Web services. By applying the content similarity module to extract the semantic similarity information between the Mashup and Web services, the two modules are seamlessly integrated into a deep neural network to accurately and quickly predict the rating information of Mashup for the Web services. Real data set on the intelligent CPS is captured and then compared with mainstream service recommendation algorithms. Experimental results show that the proposed algorithm not only efficiently completes the Web service recommendation task under the premise of sparse data but also shows better accuracy, effectivity, and privacy. Thus, the proposed method is highly suitable for the application of intelligence CPS.

Network storage services have benefited countless users worldwide due to the notable features of convenience, economy and high availability. Since a single service provider is not always reliable enough, more complex multi-cloud storage systems are developed for mitigating the data corruption risk. While a data auditing scheme is still needed in multi-cloud storage to help users confirm the integrity of their outsourced data. Unfortunately, most of the corresponding schemes rely on trusted institutions such as the centralized third-party auditor (TPA) and the cloud service organizer, and it is difficult to identify malicious service providers after service disputes. Therefore, we present a blockchain-based multi-cloud storage data auditing scheme to protect data integrity and accurately arbitrate service disputes. We not only introduce the blockchain to record the interactions among users, service providers, and organizers in data auditing process as evidence, but also employ the smart contract to detect service dispute, so as to enforce the untrusted organizer to honestly identify malicious service providers. We also use the blockchain network and homomorphic verifiable tags to achieve the low-cost batch verification without TPA. Theoretical analyses and experiments reveal that the scheme is effective in multi-cloud environments and the cost is acceptable.

Explosives combustion is primarily classified into conductive and convective combustion. In situations where confinement is sufficiently strong, the instantaneous high pressure generated by convective combustion in cracks can cause rapid fragmentation of the explosive matrix, resulting in a significant increase in the combustion surface area and triggering a high-intensity reaction with potentially catastrophic consequences. Therefore, the study of convective combustion in cracks is crucial for ensuring the safety of weapons and explosives. Previous simulation studies have primarily used finite element analysis software, which has excellent performance in handling explosive detonation processes. However, its accuracy in describing gas behavior between explosives and constrained containers is limited. This study divides the combustion process of a pre-cracked explosive in a confined space into four stages based on reasonable assumptions and simplifications. We developed a simulation method that combines the Arrhenius formula with the MWSD model to model the combustion rate of the explosive. By introducing a correction coefficient, Con, to the Arrhenius formula, the formula and MWSD model control the first and third stages of explosive combustion, respectively, while smoothly transitioning during the second stage. We used this method to numerically simulate the experimental results of Shang Hailin et al. on a crack width of 50 μm. The simulation results include the temperature field and pressure field of the first three stages of explosive combustion and the pressure rise curve of the pressure measurement point at the same location, as in the experiment. The simulation results are consistent with the experimental results.

The societal effects of industrial pollution have spurred heated debates, but more research into the influence and internal mechanism of innovative pilot city policies (IPCPs) on industrial emissions is needed. Using panel data from 276 Chinese cities between 2004 and 2018, the study employs a multi-period difference-in-differences model to explore the effects and mechanisms of pilot policies on industrial SO2 emissions, with a specific focus on how they can be mitigated by innovative techniques. The results indicate that (1) the Chinese innovative pilot city policies (CIPCPs) significantly reduced emissions in urban areas; (2) the concentration of talent, innovation policy, venture capital, and technology plays a pivotal role; and (3) the SO2 reduction effects are more pronounced in larger cities, such as super-large, mega-, and first-tier cities in the southeast, and in cities with a high market potential. This study provides empirical evidence to support the promotion of sustainable economic and social development, the resolution of environmental pollution problems, and the enhancement of public health.

Female breasts are regarded as a factor reflecting women’s morphological beauty. An appropriate bra can fulfill aesthetic needs, thus boosting self-esteem. This study proposed a method to analyze young women’s breast-bra morphological variations between two identical bras with different bra cup thicknesses. The 3D surface scan data of 129 female students who were braless and wore a thin bra (13 mm) and a thick bra (23 mm) were analyzed. Integral sections of the breasts and bra were cut at a fixed thickness of 10 mm, and slice maps were derived. Morphological parameters were extracted in braless and the two bra conditions. The variations in breast-bra shape caused by different thicknesses of bra cups were evaluated by quantifying breast ptosis, gathering, and breast slice area. The results showed that the thin bra lifted the breasts by 2.16 cm, whereas the thick bra decreased breast separation, gathering the breasts and moving them 2.15 cm laterally towards the center of the chest wall. Moreover, prediction models constructed using the critical morphological parameters were used to characterize breast-bra shape after wearing the provided bras. The findings lay the groundwork for quantifying the breast-bra shape variation caused by different bra cup thicknesses, allowing young females to choose optimally fitting bras to achieve their desired breast aesthetics.

The application of smart autonomous multi-robot systems has received increasing attention and dramatically developed, and this situation greatly promotes the development of the Internet of Things (IoT) and brings the IoT into a new stage. However, the IoT has some emerging issues, such as security and privacy; among many concerns, trusted identity authentication and consensus protocols are the main ones. To address these issues, this study proposes a novel scheme for the identity authentication and consensus algorithm . In the proposed scheme, an identity-based authentication model is designed, and all communication nodes utilize this model to establish connection and conduct data exchange. Meanwhile, a hash pool-based joint consensus algorithm is constructed. All transmission data are strictly protected by permuting hash functions from hash pool and utilizing the generated random number. This feature greatly enhances the security for the IoT. Analysis and experimental results demonstrate that the proposed scheme provides enhanced security and reliability for the IoT. Furthermore, the time efficiency is encouraging, and the communication and storage cost is promising.

The emergence of the beyond 5G (B5G) mobile networks has provided us with a variety of services and enriched our lives. The B5G super-heterogeneous network systems and highly differentiated application scenarios require highly elastic and endogenous information security, including network trust, security, and privacy. However, security issues have also been raised, which greatly threaten our information security and privacy. For example, malwares use domain generation algorithms (DGAs) to generate huge quantities of domain names and then induce users to access to steal private information, which greatly threatens our information security. In this paper, we propose an approach to detect the malicious domain name by extracting and analyzing the features using deep neural network. Unlike traditional algorithms that are generally built on tedious feature engineering, our paper utilizes the hierarchy of bidirectional recurrent neural networks (HBiRNN) to extract effective semantic features instead of traditional methods. We use the discriminator based on HBiRNN (D-HBiRNN) to detect malicious websites. This experiment verifies the validity of the algorithm and compares it with the traditional algorithm based on feature engineering. Moreover, the superiority of the algorithm is proved.

With the proliferation of IoMT (Internet of Medical Things), billions of connected medical devices are constantly producing oceans of time series sensor data, dubbed as time series for short. Considering these time series reflect various functional states of the human body, how to effectively detect the corresponding abnormalities is of great significance for smart healthcare. Accordingly, we develop a horizontal visibility graph-based temporal classification model for disease diagnosis. We conduct extensive comparison experiments on the benchmark datasets to justify the superiority of our method in term of accuracy and efficiency. Besides, we have released the codes and parameters to facilitate the community research. We propose an identifiable temporal feature selection via horizontal visibility graph for time series classification (TSC) based disease diagnosis. We conduct comparison experiments on the benchmark datasets to justify the superiority of our method in term of accuracy and efficiency. As a side contribution, we have released the codes and parameters to facilitate the community research ( https://github.com/sdujicun/SSVG ).

Along with the coming of industry 4.0 era, industrial internet of things (IIoT) plays a vital role in advanced manufacturing. It can not only connect all equipment and applications in manufacturing processes closely, but also provide oceans of sensor data for real-time work-in-process monitoring. Considering the corresponding abnormalities existing in these sensor data sequences, how to effectively implement temporal anomaly detection is of great significance for smart manufacturing. Therefore, in this paper, we proposed a novel time series anomaly detection method, which can effectively recognize corresponding abnormalities within the given time series sequences by standing on the hierarchical temporal representation. Extensive comparison experiments on the benchmark datasets have been conducted to demonstrate the superiority of our method in term of detection accuracy and efficiency on IIOT-enabled manufacturing.

Video moment localization, as an important branch of video content analysis, has attracted extensive attention in recent years. However, it is still in its infancy due to the following challenges: cross-modal semantic alignment and localization efficiency. To address these impediments, we present a cross-modal semantic alignment network. To be specific, we first design a video encoder to generate moment candidates, learn their representations, as well as model their semantic relevance. Meanwhile, we design a query encoder for diverse query intention understanding. Thereafter, we introduce a multi-granularity interaction module to deeply explore the semantic correlation between multi-modalities. Thereby, we can effectively complete target moment localization via sufficient cross-modal semantic understanding. Moreover, we introduce a semantic pruning strategy to reduce cross-modal retrieval overhead, improving localization efficiency. Experimental results on two benchmark datasets have justified the superiority of our model over several state-of-the-art competitors.

In recent years, erasure codes have become the de facto standard for data protection in large scale distributed cloud storage systems at the cost of an affordable storage overhead. However, traditional erasure coding schemes, such as Reed-Solomon codes, suffer from high reconstruction cost and I/Os. The recent past has seen a plethora of efforts to optimize the tradeoff between the reconstruction cost, I/Os and storage overhead. Quiet different from all prior studies, in this paper, our erasure coding technique makes the first attempt to take advantage of the unequal failure rates across the disks/nodes to optimize the system reliability and reconstruction performance. Specifically, our proposed technique, the Unequal Failure Protection based Local Reconstruction Code (UFP-LRC) divides the data blocks into several unequal-sized groups with local parities, assigning the data blocks stored on more failure-prone disks/nodes into the smaller-sized group, so as to provide unequal failure protection for each group. In this way, by exploiting the nonuniform local parity degrees, the proposed UFP-LRC enables the data blocks that are stored on more failure-prone disks/nodes to tolerate a greater number of failures while suffering from less repair cost than others, leading to a substantial improvement of the overall reliability and repair performance for cloud storage systems. We perform numerical analysis and build a prototype storage system to verify our approach. The analytical results show that the UFP-LRC technique gradually outperforms LRC along the increase of failure rate ratio. Also, extensive experiments show that, when compared to LRC, UFP-LRC is able to achieve a 10 to 15 percent improvement in throughput, and an 8 to 12 percent reduction in decoding latency, while retaining a comparable overall reliability.

Recently, with the rapid expansion of cities, optimizing taxi driving routes for improving taxi revenue efficiency has become the core issue of taxi system. However, most current research focuses on increasing platform revenue instead of improving drivers’ revenue in a centralized dispatch taxi system just like DiDi, which results in a slower driver income growth and greater difficulties for recruiting drivers. To solve this problem, we propose a strategy of deep reinforcement learning based on driver mode. Firstly, the sequence selection process of drivers is modeled as markov decision-making process in driver mode. Then, we propose a learning scheme based on deep Q network to optimize the driver's decision-making strategy. We know that the real selection of historical taxi drivers is very helpful to the selection of current taxi drivers, so we choose the historical record of the current location as the edge data to update the edge network. Finally, we used a real data set generated by more than 1,400 taxis in Changsha. The simulation experiments show that our scheme reduced cruising time of taxis and improved the driver's income by 4-5%. The carbon emissions are obviously reduced by saving almost 6% fuel consumption, which contributes significantly to green mobility.

Recent years have witnessed the exponential growth of time series data as the popularity of sensing devices and development of IoT techniques; time series classification has been considered as one of the most challenging studies in time series data mining, attracting great interest over the last two decades. According to the empirical evidences, temporal representation learning-based time series classification has more superiority of accuracy, efficiency and interpretability as compared to hundreds of existing time series classification methods. However, due to the high time complexity of feature process, the performance of these methods has been severely restricted. In this paper, we first presented an efficient shapelet transformation method to improve the overall efficiency of time series classification, and then, we further developed a novel enhanced recurrent neural network model for deep representation learning to further improve the classification accuracy. Experimental results on typical real-world datasets have justified the superiority of our models over several shallow and deep representation learning competitors.

Visual Question Answering (VQA) is a vital yet challenging task in the field of multimedia comprehension. In order to correctly answer questions about an image, a VQA model requires to sufficiently understand the visual scene, especially the vision-semantic reasonings between the two modalities. Traditional relation-based methods allow to encode the pairwise relations of objects to boost the VQA model performance. However, this simple strategy is deficient to exploit the abundant concepts expressed by the composition of diverse image objects, leading to sub-optimal performance. In this paper, we propose a focal and composed vision-semantic modeling method, which is a trainable end-to-end model, for better vision-semantic redundancy removal and compositionality modeling. Concretely, we first introduce the LENA cell, a plug-and-play reasoning module, which removes redundant semantic by a focal mechanism in the first step, followed by the vision-semantic compositionality modeling for better visual reasoning. We then incorporate the cell into a full LENA network, which progressively refines multimodal composed representations, and can be leveraged to infer the high-order vision-semantic in a multi-step learning way. Extensive experiments on two benchmark datasets, i.e., VQA v2 and VQA-CP v2, verify the superiority of our model as compared with several state-of-the-art baselines.

The compact and high-performance resonant tunneling diode oscillator has great potential in terahertz applications. In this work, we report a 1 THz resonant tunneling diode (RTD) oscillator with wide frequency tuning range from 935 GHz to 1.035 THz (100 GHz) under the voltage change of 0.5-0.62 V within its negative differential conductance (NDC) range, which is due to the improvement of RTD heterojunction structure.

The evolution of IoT has increased the popularity of all types of sensing devices in a variety of industrial fields and has resulted in enormous growth in the volume of sensor data. Considering the high volume and dimensionality of sensor data, the ability to perform in-depth data analysis and data mining tasks directly on the raw time series sensor data is limited. To solve this problem, we propose a novel dimensional reduction and multi-resolution representation approach for time series sensor data. This approach utilizes an appropriate number of important data points (IDPs) within a certain time series sensor data to produce a corresponding multi-resolution piecewise linear representation (MPLR), called MPLR-IDP. The results of the theoretical analyses and experiments show that MPLR-IDP can reduce the dimensionality while maintaining the important characteristics of time series data. MPLR-IDP can represent the data in a more flexible way to meet diverse needs of different users.

Along with the sustainable and rapid accumulation of user-generated contents in social networking websites, how to push a certain content to the corresponding interested users, named recommendation for short, has successfully received wide attention. Considering the continuous updated contents and the constant changing of users’ interests, recommendation is expected to be completed immediately to send the most fresh content to appropriate users after the corresponding new contents have become available. In other words, recommendation needs to meet real-time requirements to fit the content-consumption behavior of users. In the traditional recommendation system field, the corresponding attributes of users and contents could be characterized by feature vectors in a certain high-dimensional space, subsequently the recommendation problem could also be converted into how to obtain the K appropriate contents for each user, which could be called kNN join. Due to the massive, high-dimensional, and continuously updated contents, the corresponding recommendation based on traditional kNN join (continuously updating the kNN join results) will be undoubtedly faced with unacceptable computational costs. In this paper, we propose a locality-sensitive hashing (LSH)–based index called LSHI, which is built on user set to find the specific users who might be affected by the updated contents efficiently. With the help of LSHI, the recommendation lists of the affected users could be adjusted accordingly and the holistic effectiveness of the recommendation (for all users) could be guaranteed simultaneously. Finally, extensive experiments have been conducted to demonstrate the superiorities of our proposed method in this paper.

Acute wounds are worldwide problems affecting millions of people and causing heavy economic burden to national healthcare systems. Herein, we describe novel wound dressing materials relying on core/shell electrospun mats incorporated with flurbiprofen and vancomycin for achieving programmable release of anti-inflammatory and anti-bacterial agents. The shell matrix of nanofibers consisted of polyethylene oxide while the core matrix was made from a blend of silk and collagen. Several optimal mat architectures were engineered with distinct configurations, of which release profiles displayed an exponential trend, which indicates a first-order process following Fickian diffusion behavior. The flurbiprofen release lasted from 2 to 6 days, which was much faster compared to the one of vancomycin prolonged up to about 20 days. Mechanical data indicated tensile modulus, tensile strength, elongation before break of core/shell electrospun mats became enhanced or comparable to those for human skin after methanol vapor treatment. Desirable release kinetics and mechanical characteristics achieved by novel core/shell electrospun mats were attributable to induced enrichment of β-sheet phase in silk via methanol vapor treatment as well as water annealing process with time and judicious selections for matrix materials and mat configurations. The design principles considered in this study successfully addressed a range of inflammation and infection requirements in wound healing, potentially guiding construction of other biomedical coatings and devices.

Nowadays, with the proliferation of IoT (Internet of Things), we have gradually entered into a new IoE (Internet of Everything) era, in which billions of connected devices in widespread fields are constantly producing oceans of streaming time series. In order to conduct in-depth data mining researches (similarity searching, classification, clustering, prediction, etc.) based on streaming time series efficiently and effectively, time series representation should be done as the first step. In this paper, we propose a novel multi-resolution hybrid representation approach for streaming time series, which can not only generate different types of representation results in a more flexible way to meet diverse needs of users, but also be utilized as a useful preprocessing tool for the subsequent time series data mining researches. Extensive experiments on different kinds of typical time series datasets have been conducted to demonstrate the superiorities of our method.

Inflammatory reaction and cell apoptosis are two important processes in intestinal ischemia/reperfusion (II/R) injury, and exploration of effective lead compounds against II/R injury via regulating inflammation and apoptosis is critical important. In this paper, the results indicated that dioscin significantly increased cell viability, and inhibited inflammation and apoptosis caused by hypoxia-reoxygenation (H/R) injury in IEC-6 cells. in vivo II/R injury, dioscin markedly suppressed inflamma- tion and apoptosis, improved pathological changes, and depressed chiu' score in rats. Mechanistic studies indicated that dioscin notably up-regulated the expression level of MAPK13 through decreasing miR-351-5p level, and thereby decreased the expression levels of p-PKD1, NF-κB, Apaf-1, cleaved Caspase-3 and cleaved Caspase-9. Furthermore, miR-351-5p mimic and inhibitor experiments in IEC-6 cells further proved that dioscin up-regulated MAPK13 expression by decreasing miR-351-5p level to inhibit inflammation and apoptosis. Therefore, dioscin showed protective effect against II/R injury via adjusting miR-351-5/MAPK13-mediated inflammation and apoptosis. Dioscin should be considered as one potent candidate and miR-351-5/ MAPK13 should be one effective drug target for the treatment of II/R injury.

Field-effect mobility ( ${\mu }_{\mathrm{ FET}}$ ) of organic semiconductor films plays a key role in the performance of field-effect transistors (FETs). Numerical extraction of ${\mu }_{\mathrm{ FET}}$ from organic FET characteristics is not only time-consuming due to patterning of source/drain electrodes, but also frequently unreliable because of the contact resistances ( ${R} _{\mathrm{ C}}$ ) between source/drain electrodes and semiconductors. Here, we propose an approach to rapidly evaluate ${\mu }$ by a point-contact four-probes method ( $\mu _{\mathrm{ PFP}}$ ) for organic semiconductor films. Four tip-like probes quickly contact the semiconductor film surface directly, without deposition of the conventional source/drain electrodes, to simultaneously inject current and measure the electric potential. The charge density and thus conductance of the film is manipulated upon scanning gate voltage, from which the extraction of ${\mu } _{\mathrm{ PFP}}$ , in good agreement with ${\mu }_{\mathrm{ FET}}$ , could be realized in a few seconds. This method, with easily accessible setup and numerical model, substantially accelerates the evaluation of ${\mu }_{\mathrm{ PFP}}$ , and thus could help screen materials and optimize film morphology for organic FETs applications.

Time series classification (TSC) has attracted significant interest over the past decade. A lot of TSC methods have been proposed. Among these TSC methods, shapelet based methods are promising for they are interpretable, more accurate, and faster than other methods. For this, a lot of acceleration strategies are proposed. However, the accuracies of speedup methods are not ideal. To address these problems, an XGBoost classifier based on shapelet features (XG-SF) is proposed in this work. In XG-SF, an XGBoost classifier based on shapelet features is used to improve classification accuracy. Our experimental results demonstrate that XG-SF is faster than the state-of-the-art classifiers and the classification accuracy rate is also improved to a certain extent.

Along with the coming of Internet of Everything era, massive numbers of pervasive connected devices in various fields are continuously producing oceans of time series stream data. In order to carry out different kinds of data mining tasks (similarity search, classification, clustering, and prediction) based on streaming time series efficiently and effectively, segmentation and representation which segment a streaming time series into several subsequences and provide approximative representation for the raw data, should be done as the first step. With the virtue of solid theoretical foundations, piecewise linear representation (PLR) has been gained success in yielding more compact representation and fewer segments. However, the current state of art PLR methods have their own flaws: For one thing, most of current PLR methods focus on the guaranteed error bound instead of the holistic approximation error, which may lead to excessive fitting errors of segments and loss of factual research significance. For another, most of current PLR methods process streaming time series with some fixed criteria, which cannot provide a more flexible way to represent streaming time series. Motivated by the above analysis, we propose a novel continuous segmentation and multi-resolution representation approach based on turning points, which subdivides the streaming time series by a set of temporal feature points and represents the time series flexibly. Our method can not only generate more accurate approximation than the state-of-the-art of PLR algorithm, but also represent the streaming time series in a more flexible way to meet different needs of users. Extensive experiments on different kinds of typical time series datasets have been conducted to demonstrate the superiorities of our method.

Due to the continuous booming of surveillance and Web videos, video moment localization, as an important branch of video content analysis, has attracted wide attention from both industry and academia in recent years. It is, however, a non-trivial task due to the following challenges: temporal context modeling, intelligent moment candidate generation, as well as the necessary efficiency and scalability in practice. To address these impediments, we present a deep end-to-end cross-modal hashing network. To be specific, we first design a video encoder relying on a bidirectional temporal convolutional network to simultaneously generate moment candidates and learn their representations. Considering that the video encoder characterizes temporal contextual structures at multiple scales of time windows, we can thus obtain enhanced moment representations. As a counterpart, we design an independent query encoder towards user intention understanding. Thereafter, a cross-model hashing module is developed to project these two heterogeneous representations into a shared isomorphic Hamming space for compact hash code learning. After that, we can effectively estimate the relevance score of each “ moment-query ” pair via the Hamming distance. Besides effectiveness, our model is far more efficient and scalable since the hash codes of videos can be learned offline. Experimental results on real-world datasets have justified the superiority of our model over several state-of-the-art competitors.

Due to the booming of time series (a temporal data sequence, including continuous recorded values and timestamps), time series classification has been considered as one of the most challenging studies in time series data mining, attracting great interest from industry and academia over the last decades. However, the current time series classification methods have their own flaws. On the one hand, most existing methods focus on using traditional machine learning models for achieving classification accuracy, while ignoring the advantages of deep representation learning; on the other hand, the current deep neural network models only rely on one single deep learning model, and hence fail to improve performance effectively. In this paper, we propose an end-to-end representation learning model for time series classification. Concretely, we first utilize 1-D temporal convolution to obtain the feature representations. Secondly, we separately adopt the residual network and bidirectional long short-term memory network to achieve temporal representation reinforcement. Finally, we adopt the multi-layer perception network for final classification. Experimental results on open source benchmark datasets have justified the superiority of our model.

The manual assembly has a high proportion in industry. However, in many industrial scenarios, manual assembly in the Vision Blind Areas (VBAs) is time-consuming and challenging due to the lack of necessary visual information. This study presented a see-through Augmented Reality (AR) system to solve the problems during manual assembly in the vision blind area. This system enabled users to see the inner components of the VBAs cross the surface of mechanical products. The human hand and the mechanical part in a VBA were tracked and rendered in an AR HMD. We developed a prototype system and conducted a user study to evaluate the system usability, users’ performance and workload. The results indicated that this system was well integrated and easy to use. Moreover, participants worked with this system had a lower workload with improved performance.

The prop-based 3D virtual object manipulation method is widely used for interaction in Augmented Reality (AR) due to its convenience and flexibility. However, when the represented virtual object is different from the physical prop, the look and feel of the object are not well aligned. To address this problem, we present a dynamic finger remapping approach to creating a visuo-haptic illusion that dynamically adjusts the presented virtual hand posture to fit different sizes and shapes of virtual objects in AR. The finger movement toward a physical prop is synchronously remapped to the movement of the virtual fingers towards the corresponding virtual object. We developed a system that enables users to perceive consistent visual and tactile feedback while grasping and releasing various virtual objects represented by a physical prop. We conducted a user study to explore the effect of this visuo-haptic illusion on the perceived size of virtual objects, setting the sizes of the rendered virtual object and the physical prop as independent variables. We found that the perceived size of a virtual object varied with its rendered size in an almost linear fashion, while the physical prop size did not significantly affect the perception. We also conducted a second study to compare our system with a current prop-based method on virtual object manipulation. The results indicated that the remapped hands could effectively improve the realism and naturalness of the experience. Refereed/Peer-reviewed

This paper proposes a novel method to directly interact with physical content through virtual objects instead of directly using a physical proxy. The physical objects can be virtually triggered and then perform active reactions that conform to general physical laws and user expectations. We designed and implemented a prototype system to enable users to operate a virtual hammer to hit cups with different amounts of water inside. The cups could produce different audio and vibration feedback and water waves in response to the virtual hammer’s collisions. We conducted a pilot study and found that the system could enhance the realism and presence in AR interaction, compared with using a physical proxy.

With the rapid advancements of the artificial intelligence, machine learning, especially neural networks, have shown huge superiority over humans in image recognition, autonomous vehicles and medical diagnosis. However, its opacity and inexplicability provide many chances for malicious attackers. Recent researches have shown that neural networks are vulnerable to adversarial example (AE) attacks. In the testing stage, it fools the model by adding subtle perturbations to the original sample to misclassify the input, which poses a serious threat to safety-critical areas such as autonomous driving. In order to mitigate this threat, this paper proposes a hardware-assisted randomization method against AEs, where an approximate computing technique in hardware, voltage over-scaling (VOS), is used to randomize the training set of the model, then the processed data are used to generate multiple neural network models, finally multiple redundant models are used for the integrated classification and detection of the AEs. Various AE attacks on the proposed defense are evaluated to prove its effectiveness.

When interacting with virtual objects, we could discover some problems around us. Digital information has no physical properties, and sense organ(vision, tactile) does not match very well. These have a negative influence on the user’s interactive experience. In this paper, we propose to provide virtual objects with no physical properties with physical affordances to support tangible feedback via mechanical movement. In addition, we developed an interaction prototype system. The system could change the physical supports that are absorbed onto an electromagnet to adapt to the shape of different virtual objects and efficiently provide natural and consistent interactions when putting physical objects onto virtual objects in Augmented Reality (AR) scenarios.

Background and purpose Intestinal ischaemia-reperfusion (II/R) injury is a serious clinical problem. Here we have investigated novel mechanisms and new drug targets in II/R injury by searching for microRNAs regulating such injury. Experimental approach We used hypoxia/reoxygenation (H/R) of IEC-6 cell cultures and models of II/R models in rats and mice. Microarray assays were used to identify target miRNAs from rat intestinal. Real-time PCR, Western blot and dual luciferase reporter assays, and agomir and antagomir in vitro and in vivo were used to assess the effects of the target miRNA on II/R injury. Key results The miR-351-5p was differentially expressed in our models and it targeted MAPK13 and sirtuin-6. This miRNA reduced levels of sirtuin-6 and AMP-activated protein kinase phosphorylation, and activated forkhead box O3 (FoxO3α) phosphorylation to cause oxidative stress. Also, miR-351-5p markedly reduced MAPK13 level, activated polycystic kidney disease 1/NF-κB signal and increased NF-κB (p65). Moreover, miR-351-5p up-regulated levels of Bcl2-associated X, cytochrome c, apoptotic peptidase activating factor 1, cleaved-caspase 3 and cleaved-caspase 9 by reducing sirtuin-6 levels to promote apoptosis. In addition, miR-351-5p mimic in IEC-6 cells and agomir in mice aggravated these effects, and miR-351-5p inhibitor and antagomir in mice alleviated these actions. Conclusions and implications Our data showed that miR-351-5p aggravated II/R injury by promoting intestinal mucosal oxidative stress, inflammation and apoptosis by targeting MAPK13 and sirtuin-6.These data provide new insights into the mechanisms regulating II/R injury, and of miR-351-5p could be considered as a novel therapeutic target for such injury.