Using Bold-fMRI to detect cortical areas and visual fatigue related to stereoscopic vision

PurposeThe location of cortical areas and visual fatigue related to stereoscopic vision were explored by two types of fMRI designs, namely block stimulation and event-related stimulation. The stimulations consist of 2D/3D images and different depths of the stereoscopic 3D images.Method20 normal subjects were randomly divided into the block group and the event-related group. Blood oxygenation level-dependent functional magnetic resonance imaging (Bold-fMRI) was performed in two groups. Functional data was preprocessed and statistically analyzed by SPM8. The result was reported by REST.ResultIn the block stimulation group, compared to 2D image stimulation, 3D image stimulation results in more activated brain areas, including frontal lobes, occipital lobes and limbic lobes, especially in the frontal eye field (Brodmann Area 8, BA8) and middle occipital gyrus (BA18/19). In the event-related group, compared to 2D images, viewing 3D images causes significant activations in temporal lobe, mainly represented in BA19/13/31/37. Additionally, 3D image stimulation with the focus set at front depth can lead to the activation of more brain areas compared to the back depth, including inferior parietal lobule and posterior central gyrus.ConclusionThe formation of the stereoscopic vision requires the collaboration of more brain areas, so that viewing stereoscopic videos for a long period may result in visual fatigue; meanwhile, the front depth of field can contribute to more activated brain areas than the back depth of field. As a parameter of stereoscopic images, it is valid to state that the depth of field may affect visual fatigue.     

Using failure detection and diagnosis methods to detect dangerous evolutions of the driver behaviour

This paper deals with the application of techniques coming from reliability centred maintenance to detect time evolutions in systems where a human operator has a permanent role in the control loop. It describes the problems raised by such systems and proposes an approach of solution based on the characterisation of different running states of the system using exploratory data analysis techniques. An application of this approach is presented in the context of simulated car driving.     

辅助检测Linux驱动中漏洞的符号驱动环境

Linux系统中的驱动漏洞被证实是内核漏洞的主要来源,可以被利用导致严重的安全问题。通过系统模型、驱动与内核的交互和驱动与设备的交互这三部分的设计与实现,构建了符号驱动环境,用于辅助检测Linux驱动中的漏洞。使用符号驱动环境对两个真实的驱动进行检测,成功检测出了两个漏洞,证实了该工具的可行性。与SymDrive工具的性能相比,符号驱动环境执行速度快90%,覆盖率提高20%。     

使用信息增益方法选择分类器

为了减少表现差的个体分类器对集成器分类性能的影响,提高集成器分类效果及稳定性,提出了基于信息增益的分类器选择方法。该方法将高维分类器空间压缩至低维分类器空间,并在该空间内学习集成器。在多个数据集上的比较实验结果表明,该方法可行,其集成性能较理想。     

酒后驾驶无线报警系统设计与实现

详细介绍了将检测技术与无线通信技术用于预防酒后驾驶系统的设计方法。系统选用半导体型酒精传感器完成对司乘人员呼气的酒精检测,经放大、比较和判断,通过编程把判断结果分别形成语音预警信号,再利用公用通信网(G网、CDMA或固定电话网)把语音预警信号传送给车主或乘客。测试结果表明,对司乘人员的饮酒情况判定准确,接收的语音信号清晰可靠。证明了该设计方法的有效性和可行性。     

Using online linear classifiers to filter spam emails

The performance of two online linear classifiers—the Perceptron and Littlestone’s Winnow—is explored for two anti-spam filtering benchmark corpora—PU1 and Ling-Spam. We study the performance for varying numbers of features, along with three different feature selection methods: information gain (IG), document frequency (DF) and odds ratio. The size of the training set and the number of training iterations are also investigated for both classifiers. The experimental results show that both the Perceptron and Winnow perform much better when using IG or DF than using odds ratio. It is further demonstrated that when using IG or DF, the classifiers are insensitive to the number of features and the number of training iterations, and not greatly sensitive to the size of training set. Winnow is shown to slightly outperform the Perceptron. It is also demonstrated that both of these online classifiers perform much better than a standard Naïve Bayes method. The theoretical and implementation computational complexity of these two classifiers are very low, and they are very easily adaptively updated. They outperform most of the published results, while being significantly easier to train and adapt. The analysis and promising experimental results indicate that the Perceptron and Winnow are two very competitive classifiers for anti-spam filtering.     

使用Fisher线性判别方法的提取分类器

为了消除个体分类器间的相关性,提高集成器分类性能及稳定性,提出了基于Fisher线性判别方法的分类器提取方法。该方法将高维分类器空间压缩至低维分类器空间,并在该空间内学习集成器。在多个数据集上的比较实验结果表明,该方法可行,其集成性能较理想。     

Resampling imbalanced data to detect fake reviews using machine learning classifiers and textual-based features

Fraudulent online sellers often collude with reviewers to garner fake reviews for their products. This act undermines the trust of buyers in product reviews, and potentially reduces the effectiveness of online markets. Being able to accurately detect fake reviews is, therefore, critical. In this study, we investigate several preprocessing and textual-based featuring methods along with machine learning classifiers, including single and ensemble models, to build a fake review detection system. Given the nature of product review data, where the number of fake reviews is far less than that of genuine reviews, we look into the results of each class in detail in addition to the overall results. We recognise from our preliminary analysis that, owing to imbalanced data, there is a high imbalance between the accuracies for different classes (e.g., 1.3% for the fake review class and 99.7% for the genuine review class), despite the overall accuracy looking promising (around 89.7%). We propose two dynamic random sampling techniques that are possible for textual-based featuring methods to solve this class imbalance problem. Our results indicate that both sampling techniques can improve the accuracy of the fake review class—for balanced datasets, the accuracies can be improved to a maximum of 84.5% and 75.6% for random under and over-sampling, respectively. However, the accuracies for genuine reviews decrease to 75% and 58.8% for random under and over-sampling, respectively. We also discover that, for smaller datasets, the Adaptive Boosting ensemble model outperforms other single classifiers; whereas, for larger datasets, the performance improvement from ensemble models is insignificant compared to the best results obtained by single classifiers.

     

Using active learning to adapt remote sensing image classifiers

The validity of training samples collected in field campaigns is crucial for the success of land use classification models. However, such samples often suffer from a sample selection bias and do not represent the variability of spectra that can be encountered in the entire image. Therefore, to maximize classification performance, one must perform adaptation of the first model to the new data distribution. In this paper, we propose to perform adaptation by sampling new training examples in unknown areas of the image. Our goal is to select these pixels in an intelligent fashion that minimizes their number and maximizes their information content. Two strategies based on uncertainty and clustering of the data space are considered to perform active selection. Experiments on urban and agricultural images show the great potential of the proposed strategy to perform model adaptation.     

Using multiple and negative target rules to make classifiers more understandable

One major goal for data mining is to understand data. Rule based methods are better than other methods in making mining results comprehensible. However, current rule based classifiers make use of a small number of rules and a default prediction to build a concise predictive model. This reduces the explanatory ability of the rule based classifier. In this paper, we propose to use multiple and negative target rules to improve explanatory ability of rule based classifiers. We show experimentally that this understandability is not at the cost of accuracy of rule based classifiers.     

The architecture and performance of the face and eyes detection system based on the Haar cascade classifiers

The precise face and eyes detection is essential in many human–machine interface systems. Therefore, it is necessary to develop a reliable and efficient object detection method. In this paper we present the architecture of a hierarchical face and eyes detection system using the Haar cascade classifiers (HCC) augmented with some simple knowledge-based rules. The influence of the training procedure on the performance of the particular HCCs has been investigated. Additionally, we compared the efficiency of other authors’ face and eyes HCCs with the efficiency of those trained by us. By applying the proposed system to the set of 10,000 test images we were able to properly detect and precisely localize 94% of the eyes.     

Using string matching to detect video transitions

The detection of shot boundaries in videos captures the structure of the image sequences by the identification of transitional effects. This task is important in the video indexing and retrieval domain. The video slice or visual rhythm is a single two-dimensional image sampling that has been used to detect several types of video events, including transitions. We use the longest common subsequence (LCS) between two strings to transform the video slice into one-dimensional signals obtaining a highly simplified representation of the video content. We also developed a chain of mathematical morphology operations over these signals leading to the detection of the most frequent video transitions, namely, cut, fade, and wipe. The algorithms are tested with success with various genres of videos.     

Using measurement-driven modeling to provide empirical feedback to software developers

Several authors have explored the application of classification methods to software development. These studies have concentrated on identifying modules that are difficult to develop or that have high fault density. While this information is important, it provides little help in determining appropriate corrective action. This article extends previous work by applying one classification method, classification tree analysis (CTA), to more a fine-grained problem routinely encountered by developers. In this article, we use CTA to identify software modules that have specific types of faults (e.g., logic, interface, etc.) We evaluate this approach using data collected from six actual software projects. Overall, CTA was able to correctly differentiate faulty modules from fault-free modules in 72% of cases. Furthermore, 82% of the faulty modules were correctly identified. We also show that CTA outperformed two simpler classification strategies.     

基于云计算的多特征疲劳驾驶检测系统研究与设计

随着私家车的普及,人们对汽车安全性、舒适性要求不断提高,通过对当前车载系统分析和汽车驾驶员疲劳驾驶状态研究,提出了一种基于信息融合的多特征疲劳驾驶检测方案。方案采用高性能嵌入式系统平台与云计算相结合的方式,首先,通过嵌入式系统采集驾驶员面部图像;然后,将数据传输到Face+ 云计算平台,分析当前驾驶人员身份、年龄与微笑程度;最后,采用数字图像处理技术计算驾驶员头部位移以及统计眼睛眨动规律,综合三种指标预测驾驶员是否处于疲劳状态,实时监测驾驶员驾驶全过程。当检测到驾驶员处于疲劳驾驶状态,则通过语音的方式提醒驾驶员注意行车安全、谨慎驾驶。测试结果表明：该方案检测精度高、实时性强,并且易于和车载系统整合并推广使用。     

Using physiological signals to detect natural interactive behavior

Many researchers in the Human Robot Interaction (HRI) and Embodied Conversational Agents (ECA) domains try to build robots and agents that exhibit human-like behavior in real-world close encounter situations. One major requirement for comparing such robots and agents is to have an objective quantitative metric for measuring naturalness in various kinds of interactions. Some researchers have already suggested techniques for measuring stress level, awareness etc using physiological signals like Galvanic Skin Response (GSR) and Blood Volume Pulse (BVP). One problem of available techniques is that they are only tested with extreme situations and cannot according to the analysis provided in this paper distinguish the response of human subjects in natural interaction situations. One other problem of the available techniques is that most of them require calibration and some times ad-hoc adjustment for every subject. This paper explores the usefulness of various kinds of physiological signals and statistics in distinguishing natural and unnatural partner behavior in a close encounter situation. The paper also explores the usefulness of these statistics in various time slots of the interaction. Based on this analysis a regressor was designed to measure naturalness in close encounter situations and was evaluated using human-human and human-robot interactions and shown to achieve statistically significant distinction between natural and unnatural situations.     

Using compiler optimization techniques to detect equivalent mutants

Mutation analysis is a software testing technique that requires the tester to generate test data that will find specific, well-defined errors. Mutation testing executes many slightly differing versions, called mutants, of the same program to evaluate the quality of the data used to test the program. Although these mutants are generated and executed efficiently by automated methods, many of the mutants are functionally equivalent to the original program and are not useful for testing. Recognizing and eliminating equivalent mutants is currently done by hand, a time-consuming and arduous task. This problem is currently a major obstacle to the practical application of mutation testing. This paper presents extensions to previous work in detecting equivalent mutants; specifically, algorithms for determining several classes of equivalent mutants are presented, an implementation of these algorithms is discussed, and results from using this implementation are presented. These algorithms are based on data flow analysis and six compiler optimization techniques. Each of these techniques is described together with how they are used to detect equivalent mutants. The design of the tool and some experimental results using it are also presented.     

Using static symbolic execution to detect buffer overflows

Buffer overrun remains one of the main sources of errors and vulnerabilities in the C/C++ source code. To detect such kind of defects, static analysis is widely used. In this paper, we propose a path-sensitive static analysis based on symbolic execution with state merging. For buffers with compile-time-known sizes, we present an interprocedural path- and context-sensitive overrun detection algorithm that finds program points satisfying a proposed error definition. The described approach was implemented in the Svace static analyzer without significant loss of performance. On Android 5.0.2, these detectors generated 351 warnings, 64% of which were true positives. In addition, we describe a prototype of an intraprocedural heap buffer overflow detector and present an example of a defect found by this detector.     

Applying neural network analysis on heart rate variability data to assess driver fatigue

Long duration driving is a significant cause of fatigue related accidents on motorways. Fatigue caused by driving for extended hours can acutely impair driver’s alertness and performance. This papers presents an artificial intelligence based system which could detect early onset of fatigue in drivers using heart rate variability (HRV) as the human physiological measure. The detection performance of neural network was tested using a set of electrocardiogram (ECG) data recorded under laboratory conditions. The neural network gave an accuracy of 90%. This HRV based fatigue detection technique can be used as a fatigue countermeasure.     

Using artificial anomalies to detect unknown and known network intrusions

Intrusion detection systems (IDSs) must be capable of detecting new and unknown attacks, or anomalies. We study the problem of building detection models for both pure anomaly detection and combined misuse and anomaly detection (i.e., detection of both known and unknown intrusions). We show the necessity of artificial anomalies by discussing the failure to use conventional inductive learning methods to detect anomalies. We propose an algorithm to generate artificial anomalies to coerce the inductive learner into discovering an accurate boundary between known classes (normal connections and known intrusions) and anomalies. Empirical studies show that our pure anomaly-detection model trained using normal and artificial anomalies is capable of detecting more than 77% of all unknown intrusion classes with more than 50% accuracy per intrusion class. The combined misuse and anomaly-detection models are as accurate as a pure misuse detection model in detecting known intrusions and are capable of detecting at least 50% of unknown intrusion classes with accuracy measurements between 75 and 100% per class.