Search

Detection of genomics rearrangements is a tough task, because of the size of data to be processed. As genome sequences may consist of hundreds of millions symbols, it is not only practically impossible to compare them by hand, but it is also complex problem for computer software. The way to significantly accelerate the process is to use rearrangement detection algorithm based on unique short sequences called markers. The algorithm described in this paper develops markers using base genome and find the markers positions on other genome. The algorithm has been extended by support for ambiguity symbols. Web application with graphical user interface has been created using three-layer architecture, where users could run the task simultaneously. The accuracy and efficiency of proposed solution has been studied using generated and real data.

In the work the hybrid expert system for stroke diagnosis was presented. The base of expert system consists of neural network and production rules. This program can quickly and accurately set to the patient preliminary and final diagnoses, get examination and treatment plans, print data of patient, analyze statistics data and perform parameterized search for patients.

Human Epithelial-2 (HEp-2) cell images staining patterns classification have been widely used to identify autoimmune diseases by the anti-Nuclear antibodies (ANA) test in the Indirect Immunofluorescence (IIF) protocol. Because manual test is time consuming, subjective and labor intensive, image-based Computer Aided Diagnosis (CAD) systems for HEp-2 cell classification are developing. However, methods proposed recently are mostly manual features extraction with low accuracy. Besides, the scale of available benchmark datasets is small, which does not exactly suitable for using deep learning methods. This issue will influence the accuracy of cell classification directly even after data augmentation. To address these issues, this paper presents a high accuracy automatic HEp-2 cell classification method with small datasets, by utilizing very deep convolutional networks (VGGNet). Specifically, the proposed method consists of three main phases, namely image preprocessing, feature extraction and classification. Moreover, an improved VGGNet is presented to address the challenges of small-scale datasets. Experimental results over two benchmark datasets demonstrate that the proposed method achieves superior performance in terms of accuracy compared with existing methods.

Multiple-instance learning (MIL) has been successfully utilized in image retrieval. Existing approaches cannot select positive instances correctly from positive bags which may result in a low accuracy. In this paper, we propose a new image retrieval approach called multiple instance learning based on instance-consistency (MILIC) to mitigate such issue. First, we select potential positive instances effectively in each positive bag by ranking instance-consistency (IC) values of instances. Then, we design a feature representation scheme, which can represent the relationship among bags and instances, based on potential positive instances to convert a bag into a single instance. Finally, we can use a standard single-instance learning strategy, such as the support vector machine, for performing object-based image retrieval. Experimental results on two challenging data sets show the effectiveness of our proposal in terms of accuracy and run time.

Raw scanned 3D point clouds are usually irregularly distributed due to the essential shortcomings of laser sensors, which therefore poses a great challenge for high-quality 3D surface reconstruction. This paper tackles this problem by proposing a local hierarchical clustering (LHC) method to improve the consistency of point distribution. Specifically, LHC consists of two steps: 1) adaptive octree-based decomposition of 3D space, and 2) hierarchical clustering. The former aims at reducing the computational complexity and the latter transforms the non-uniform point set into uniform one. Experimental results on real-world scanned point clouds validate the effectiveness of our method from both qualitative and quantitative aspects.

The watermarking technique can be used to protect the copyright of relational databases by hiding the ownership information into the relational databases. Difference expansion (DE) technique is one of the common reversible watermarking techniques for numerical relational databases. However, most previous schemes based on DE suffer the problem of low embedding capacity when the difference values between different attributes are relatively large. In this paper, we propose a novel reversible watermarking scheme to solve the above problem. In the scheme, a mapping difference expansion (MDE) method is proposed to convert the differences between attributes to small mapping differences. Based on the MDE, an attribute and tuple selection algorithm is designed to select the suitable data for watermarking, which can increase embedding capacity and reduce distortion. In addition, the majority voting technique is utilized to enhance the robustness of watermarking with the high embedding capacity. The experimental results have shown that the proposed scheme can provide higher embedding capacity, lower distortion and stronger robustness than other schemes.

In this paper, we propose a framework for low cost secure electronic voting system based on face recognition. Essentially Local Binary Pattern (LBP) is used for face feature characterization in texture format followed by chi-square distribution is used for image classification. Two parallel systems are developed based on smart phone and web applications for face learning and verification modules. The proposed system has two tire security levels by using person ID followed by face verification. Essentially class specific threshold is associated for controlling the security level of face verification. Our system is evaluated three standard databases and one real home based database and achieve the satisfactory recognition accuracies. Consequently our propose system provides secure, hassle free voting system and less intrusive compare with other biometrics.

Due to its capability of capturing the kinematic properties of a target object, radar micro-Doppler signatures (m-DS) play an important role in radar target classification. This is particularly evident from the remarkable number of research papers published every year on m-DS for various applications. However, most of these works rely on the support vector machine (SVM) for target classification. It is well known that training an SVM is computationally expensive due to its nature of search to locate the supporting vectors. In this paper, the classifier learning problem is addressed by a total error rate (TER) minimization where an analytic solution is available. This largely reduces the search time in the learning phase. The analytically obtained TER solution is globally optimal with respect to the classification total error count rate. Moreover, our empirical results show that TER outperforms SVM in terms of classification accuracy and computational efficiency on a five-category radar classification problem.

Strong Jumping Emerging Patterns (SJEPs) are data mining patterns which have strong discriminating abilities in classification. However, SJEPs mining algorithms in current years are usually achieved by the data structure, tree. These existing algorithms using the tree structure are difficult to achieve excellent performance. In this paper, we propose a novel method of mining SJEPs named PPSJEP. This algorithm is based on a novel data structure called NSJEP-list, which is improved from the N-list. We use the NSJEP-lists to replace the tree structure. First, we get the individual items’ NSJEP-lists from the tree. Then we use the intersection of NSJEP-lists to get the longer itemsets’ NSJEP-lists which includes the information of the position and the count in each class. And we mine the SJEPs through the information. Experiments are performed on six UCI datasets. Compared with existing algorithm in running time and classification accuracy, the experimental results show that our algorithm uses less time to mine SJEPs and get the same classification accuracy, especially in lower minimum support threshold.

Deep learning is a very hot topic currently in pattern recognition and artificial intelligence researches. Aiming at the practical problem that people usually don’t know correct classifications some rubbish should belong to, based on the powerful image classification ability of the deep learning method, we have designed a prototype system to help users to classify kinds of rubbish. Firstly the CaffeNet Model was adopted for our classification network training on the ImageNet dataset, and the trained network was deployed on a web server. Secondly an android app was developed for users to capture images of unclassified rubbish, upload images to the web server for analyzing backstage and retrieve the feedback, so that users can obtain the classification guide by an android device conveniently. Tests on our prototype system of rubbish classification show that: an image of one single type of rubbish with origin shape can be better used to judge its classification, while an image containing kinds of rubbish or rubbish with changed shape may fail to help users to decide rubbish’s classification. However, the system still shows promising auxiliary function for rubbish classification if the network training strategy can be optimized further.

The recent proposed Indexing-First-One (IFO) hashing is a latest technique that is particularly adopted for eye iris template protection, i.e. IrisCode. However, IFO employs the measure of Jaccard Similarity (JS) initiated from Min-hashing has yet been adequately discussed. In this paper, we explore the nature of JS in binary domain and further propose a mathematical formulation to generalize the usage of JS, which is subsequently verified by using CASIA v3-Interval iris database. Our study reveals that JS applied in IFO hashing is a generalized version in measure two input objects with respect to Min-Hashing where the coefficient of JS is equal to one. With this understanding, IFO hashing can propagate the useful properties of Min-hashing, i.e. similarity preservation, thus favorable for similarity searching or recognition in binary space.

Numerous fuzzy pattern mining methods have been proposed to address the uncertainty and incompleteness of quantitative data. Traditional fuzzy pattern mining methods generally have to transform the original quantitative values into either crystal items or fuzzy regions first, which is hard to apply without comprehensive domain knowledge. In addition, existing numerical pattern mining methods generally suffer high computational cost. Inspired by the above problems, we put forward an efficient maximal approximate numerical frequent pattern mining (MANFPM) method without fuzzy item or region specification. Experimental results have validated its scalability and effectiveness for application in emitter entity resolution.

This paper proposes a multi-attribute automatic optimization method for lens system design. The method is based on the exploitation of the multi-attribute property of lens systems. Quality attribute and sensitivity attribute of lens system are our main concerns in the proposed method. The quality attribute is considered to characterize the optical performance and the physical constraints of lens system. The sensitivity attribute is perceived as the description of feasibility and practicality in lens manufacturing. Modified coordinate-wise algorithm is employed to perform the optimization of imaging performance as well as specific constraints. The sensitivity attribute is processed to incorporate manufacturing sensitivity into the metrics to evaluate the system more comprehensively by means of minimizing incident and refracted ray angles on optical surfaces. As a result, compromise solutions between quality attribute and sensitivity attribute are obtained. Experiments on two typical lens designs demonstrate the effectiveness of the proposed method.

NIR technology has been available for a long time, certainly more than 50 years. Without any doubt, it has found many niche applications, especially in the pharmaceutical, food, agriculture and other industries due to its flexibility. There are a number of advantages over other existing analytical technologies we can list, for example virtually no need for sample preparation; usually NIR does not demand sample destruction and subsequent discard; NIR provides fast results; NIR does not require extensive operator training and carries small operating costs. However, the key point about NIR technology is the fact that it´s more related to statistics than chemistry or, in other words, we are more concerned about analyzing and distinguishing features within the data than looking deep into the chemical entities themselves. A simple scan reading in the NIR range usually involves huge inflows of data points. Usually we decompose the signals into hundreds of predictor variables and use complex algorithms to predict classes or quantify specific content. NIR is all about math, especially by converting chemical information into numbers. Easier said than done. A NIR signal is a very complex one. Usually the signal responses are not specific to a particular material, rather, each group´s responses add up, thus providing low specificity of a spectral reading. This paper proposes a simple and efficient method to analyze and compare NIR spectra for the purpose of identifying the presence of active pharmaceutical ingredients in finished products using low cost NIR scanning devices connected to the internet cloud.

The United States increasingly relies on cyber-physical systems to conduct military and commercial operations. Attacks on these systems have increased dramatically around the globe. The attackers constantly change their methods, making state-of-the-art commercial and military intrusion detection systems ineffective. In this paper, we present a model to identify functional behavior of network devices from netflow traces. Our model includes two innovations. First, we define novel features for a host IP using detection of application graph patterns in IP’s host graph constructed from 5-min aggregated packet flows. Second, we present the first application, to the best of our knowledge, of Graph Semi-Supervised Learning (GSSL) to the space of IP behavior classification. Using a cyber-attack dataset collected from NetFlow packet traces, we show that GSSL trained with only 20% of the data achieves higher attack detection rates than Support Vector Machines (SVM) and Naive Bayes (NB) classifiers trained with 80% of data points. We also show how to improve detection quality by filtering out web browsing data, and conclude with discussion of future research directions.

This paper outlines a framework for quantifying and optimizing the value of information from structural health monitoring (SHM) technology deployed on large infrastructure, which may sustain damage in a series of earthquakes (the main and the aftershocks). The evolution of the damage state of the infrastructure without or with SHM is presented as a time-dependent, stochastic, discrete-state, observable and controllable nonlinear dynamical system. The pre-posterior Bayesian analysis and the decision tree are used for quantifying and optimizing the value of SHM information. An optimality problem is then formulated how to decide on the adoption of SHM and how to manage optimally the usage and operations of the possibly damaged infrastructure and its repair schedule using the information from SHM. The objective function to minimize is the expected total cost or risk.

Crop improvement programs require large and meticulous selection processes that effectively and accurately collect and analyze data to generate quality plant products as efficiently as possible, develop superior cropping and/or crop improvement methods. Typically, data collection for such testing is performed by field teams using hand-held instruments or manually-controlled devices. Although steps are taken to reduce error, the data collected in such manner can be unreliable due to human error and fatigue, which reduces the ability to make accurate selection decisions. Monsanto engineering teams have developed a high-clearance mobile platform (Rover) as a step towards high throughput and high accuracy phenotyping at an industrial scale. The rovers are equipped with GPS navigation, multiple cameras and sensors and on-board computers to acquire data and compute plant vigor metrics per plot. The supporting IT systems enable automatic path planning, plot identification, image and point cloud data QA/QC and near real-time analysis where results are streamed to enterprise databases for additional statistical analysis and product advancement decisions. Since the rover program was launched in North America in 2013, the number of research plots we can analyze in a growing season has expanded dramatically. This work describes some of the successes and challenges in scaling up of the rover platform for automated phenotyping to enable science at scale.

Most existing radar algorithms are developed under the assumption that the environment, data clutter, is known and stationary. However, in practice, the characteristics of clutter can vary enormously in time depending on the operational scenarios. If unaccounted for, these nonstationary variabilities may drastically hinder the radar performance. It is essential that the radar systems dynamically detect changes in the environment, and adapt to these changes by learning the new statistical characteristics of the environment. In this paper, we employ sparse recovery for clutter identification, specifically we identify the statistical profile the clutter follows. We use Monte Carlo simulations to simulate and test clutter data coming from various distributions.

Modeling large military targets is a challenge as they can be complex systems encompassing myriad combinations of human, technological, and social elements that interact, leading to complex behaviors. Moreover, such targets have multiple components and structures, extending across multiple spatial and temporal scales, and are in a state of change, either in response to events in the environment or changes within the system. Complex adaptive system (CAS) theory can help in capturing the dynamism, interactions, and more importantly various emergent behaviors, displayed by the targets. However, a key stumbling block is incorporating information from various intelligence, surveillance and reconnaissance (ISR) sources, while dealing with the inherent uncertainty, incompleteness and time criticality of real world information. To overcome these challenges, we present a probabilistic reasoning network based framework called complex adaptive Bayesian Knowledge Base (caBKB). caBKB is a rigorous, overarching and axiomatic framework that models two key processes, namely information aggregation and information composition. While information aggregation deals with the union, merger and concatenation of information and takes into account issues such as source reliability and information inconsistencies, information composition focuses on combining information components where such components may have well defined operations. Since caBKBs can explicitly model the relationships between information pieces at various scales, it provides unique capabilities such as the ability to de-aggregate and de-compose information for detailed analysis. Using a scenario from the Network Centric Operations (NCO) domain, we will describe how our framework can be used for modeling targets with a focus on methodologies for quantifying NCO performance metrics.

Singapore’s “smart city” agenda is driving the government to provide public access to a broader variety of urban informatics sources, such as images from traffic cameras and information about buses servicing different bus stops. Such informatics data serves as probes of evolving conditions at different spatiotemporal scales. This paper explores how such multi-modal informatics data can be used to establish the normal operating conditions at different city locations, and then apply appropriate outlier-based analysis techniques to identify anomalous events at these selected locations. We will introduce the overall architecture of sociophysical analytics, where such infrastructural data sources can be combined with social media analytics to not only detect such anomalous events, but also localize and explain them. Using the annual Formula-1 race as our candidate event, we demonstrate a key difference between the discriminative capabilities of different sensing modes: while social media streams provide discriminative signals during or prior to the occurrence of such an event, urban informatics data can often reveal patterns that have higher persistence, including before and after the event. In particular, we shall demonstrate how combining data from (i) publicly available Tweets, (ii) crowd levels aboard buses, and (iii) traffic cameras can help identify the Formula-1 driven anomalies, across different spatiotemporal boundaries.

The Open Standard for Unattended Sensors (OSUS) was developed by DIA and ARL to provide a plug-n-play platform for sensor interoperability. Our objective is to use the standardized data produced by OSUS in performing data analytics on information obtained from various sensors. Data analytics can be integrated in one of three ways: within an asset itself; as an independent plug-in designed for one type of asset (i.e. camera or seismic sensor); or as an independent plug-in designed to incorporate data from multiple assets. As a proof-of-concept, we develop a model that can be used in the second of these types – an independent component for camera images. The dataset used was collected as part of a demonstration and test of OSUS capabilities. The image data includes images of empty outdoor scenes and scenes with human or vehicle activity. We design, test, and train a convolution neural network (CNN) to analyze these images and assess the presence of activity in the image. The resulting classifier labels input images as empty or activity with 86.93% accuracy, demonstrating the promising opportunities for deep learning, machine learning, and predictive analytics as an extension of OSUS’s already robust suite of capabilities.

The U.S. Army RDECOM CERDEC NVESD MSD’s target acquisition models have been used for many years by the military analysis community for sensor design, trade studies, and field performance prediction. This paper analyzes the results of perception tests performed to compare the results of a field DRI (Detection, Recognition, and Identification Test) performed in 2009 to current Soldier performance viewing the same imagery in a laboratory environment and simulated imagery of the same data set. The purpose of the experiment is to build a robust data set for use in the virtual prototyping of infrared sensors. This data set will provide a strong foundation relating, model predictions, field DRI results and simulated imagery.

Detection and classification of targets is a process that we perform every day, often without even realizing it; the ability to locate and recognize a specific face in a crowd, to be able to identify a bird by its song, or to know what is for dinner simply by the aroma. Much effort has been spent in trying to replicate these uncanny abilities of our senses through purpose-built sensors. For many of these sensing modalities, automatic target recognition algorithms have been developed to automate the process of locating and recognizing the occurrence of a specific state from within the sensed data. This survey is an overview of the metrics used to assess and quantify the quality or ‘goodness’ of the data, mainly for imaging sensors, from the perspective of automatic target recognition performance. Digital image data go through several transformations as they are moved from the recording sensor to personal electronic devices, resulting in distortions. There has been considerable research in the area of quality assessment of digital image and video media. Often, the image quality assessment approaches rely on having the original image or at least information about the original image available for the evaluation. These approaches may also rely on knowing the source of distortion. The accuracy is typically gauged using the image quality assessment made by the human visual system. As a consequence, the metrics measure aspects of the image that are tied closely to the human perception of quality. Evaluating image quality from an automatic target recognition perspective offers many challenges, as the original data is not likely to be available. Additionally, there may be multiple unknown sources of image quality degradation forcing the need for a general solution. Finally, the human perception of ‘good quality’ may not necessarily correlate to optimal image quality for automatic target recognition performance, in which case metrics modeled after the human visual system may not be the best choice for image quality assessment. The goal of this survey is to provide a high-level background and overview of how image quality assessment is currently performed for different sensing modalities. Additionally, the challenges that are being addressed within the various disciplines will be discussed and commonalities will be highlighted.

This paper describes characteristics of information flow on social channels, as a function of content type and relations among individual sources, distilled from analysis of Twitter data as well as human subject survey results. The working hypothesis is that individuals who propagate content on social media act (e.g., decide whether to relay information or not) in accordance with their understanding of the content, as well as their own beliefs and trust relations. Hence, the resulting aggregate content propagation pattern encodes the collective content interpretation of the underlying group, as well as their relations. Analysis algorithms are described to recover such relations from the observed propagation patterns as well as improve our understanding of the content itself in a language agnostic manner simply from its propagation characteristics. An example is to measure the degree of community polarization around contentious topics, identify the factions involved, and recognize their individual views on issues. The analysis is independent of the language of discourse itself, making it valuable for multilingual media, where the number of languages used may render language-specific analysis less scalable.

Search planning for UUV missions involves a complex optimization problem over multiple degrees of freedom. The detection performance of individual searches is governed by the local environment and the sensor settings that determine the rate at which target objects of interest can be identified. This paper addresses some of the computational aspects of collaborative search planning when multiple search agents seek to find hidden objects (i.e. mines) in adverse local operating environments where the detection process is prone to false alarms. For these conditions, we apply a receiver operator characteristic (ROC) analysis to model detection performance and develop a Bayesian risk objective that enumerates the required number of detections for validation as part of the modeling paradigm. In this paper we consider an expanded optimization process whereby ROC operating points are optimized simultaneously for all validation criteria with the best performing combinations selected. Details of its application within a broader search planning context are discussed. An analysis of this optimization process is presented. Numerical results are provided to demonstrate the effectiveness of the approach.

When new and unique task difficulties are requested to be determined, it is important to use methodologies that are consistent with previous research. Unfortunately, some new tasks break the paradigm of past research and require new techniques in order to properly determine their difficulty. This paper describes the process of determining the difficulty for tasks that are unique in that they have a null case (where no object or motion is present) and because these tasks have been requested to be quantified in environments that potentially contain high amounts of atmospheric turbulence. Because each of the calculated V50’s was based upon an assumption, a secondary field collection was necessary in order to validate which model assumptions correlated properly to field performance data.

In today’s battlefield environments, analysts are inundated with real-time data received from the tactical edge that must be evaluated and used for managing and modifying current missions as well as planning for future missions. This paper describes a framework that facilitates a Value of Information (VoI) based data analytics tool for information object (IO) analysis in a tactical and command and control (C2) environment, which reduces analyst work load by providing automated or analyst assisted applications. It allows the analyst to adjust parameters for data matching of the IOs that will be received and provides agents for further filtering or fusing of the incoming data. It allows for analyst enhancement and markup to be made to and/or comments to be attached to the incoming IOs, which can then be re-disseminated utilizing the VoI based dissemination service. The analyst may also adjust the underlying parameters before re-dissemination of an IO, which will subsequently adjust the value of the IO based on this new/additional information that has been added, possibly increasing the value from the original. The framework is flexible and extendable, providing an easy to use, dynamically changing Command and Control decision aid that focuses and enhances the analyst workflow.

Wideband electromagnetic induction (WEMI) sensors can be used to detect, classify, and locate metallic targets buried underground. By examining the WEMI model from a new perspective, a low-rank equivalence for the WEMI data is obtained. The low-rank physical WEMI model directly leads to a new “filterless” processing framework that differs radically from traditional approaches to WEMI processing. The new framework enables target signature recovery independent of location estimation. Processing of field data with the new filterless WEMI framework is presented to show classification results based on the low-rank target signature.

Creating entity network graphs is a manual, time consuming process for an intelligence analyst. Beyond the traditional big data problems of information overload, individuals are often referred to by multiple names and shifting titles as they advance in their organizations over time which quickly makes simple string or phonetic alignment methods for entities insufficient. Conversely, automated methods for relationship extraction and entity disambiguation typically produce questionable results with no way for users to vet results, correct mistakes or influence the algorithm’s future results. We present an entity disambiguation tool, DRADIS, which aims to bridge the gap between human-centric and machinecentric methods. DRADIS automatically extracts entities from multi-source datasets and models them as a complex set of attributes and relationships. Entities are disambiguated across the corpus using a hierarchical model executed in Spark allowing it to scale to operational sized data. Resolution results are presented to the analyst complete with sourcing information for each mention and relationship allowing analysts to quickly vet the correctness of results as well as correct mistakes. Corrected results are used by the system to refine the underlying model allowing analysts to optimize the general model to better deal with their operational data. Providing analysts with the ability to validate and correct the model to produce a system they can trust enables them to better focus their time on producing higher quality analysis products.

Real-time Analytics Platform for Interactive Data-mining (RAPID), a collaboration of University of Melbourne and Australia’s Defense Science and Technology Group (DSTG), consumes data streams, performs analytics computations, and produces high-quality knowledge for analysts. RAPID takes topic seed words and autonomously identifies emerging keywords in the data. Users direct the system, setting time-windowing parameters, thresholds, update intervals and sample rates. Apache Storm and Apache Kafka permit real-time streaming while logging options support off-line processing. Decision-support scenarios feature Commander Critical Information Requirements involving comparisons over time and time-sequencing of events, capabilities particularly well-served by RAPID technology, to be demonstrated in the presentation.

In this paper, improving data association process by increasing the probability of detecting valid data points (measurements obtained from ESM/RADAR system) in the presence of noise for location and target tracking are discussed. This develop a multisensor data association algorithm that fuses information from the multiple ESM receiver and surveillance RADAR. The develop a novel algorithm by self-organizing fuzzy neural network (SO-FNN) for multiple ESM-to-ESM (measurement-to-measurement data association) and ESMs-to-RADAR (track-to-track data association) problem in dense clutter environment. An adaptive search based on SO-FNN of the distance threshold measure is then used to detect valid filtered data point for data association. Simulation results demonstrate the effectiveness and better performance when compared to conventional algorithm. The paper is organized as follows. Section 1 is the problem formulation. Section 2 design the new data association algorithm based on SO-FNN data association system design. Section 3 describes ESM-to-ESM and ESM-to-RADAR (measurement-to-measurement data association and track-to-track data association) scenario and the simulation results are presented and discussed. The summary are drawn in section 4, respectively.

In a two class label scenario, classi…cation systems may be used to assess whether or not an element of interest belongs to the “target”or “non-target”class. The performance of the system is summarized visually as a trade- o¤ between the proportions of elements correctly labeled as “target”plotted against the proportion of elements incorrectly labeled as “target.” These proportions are empirical estimates of the true and false positive rates, and their trade-o¤ plot is known as a receiver operating characteristic (ROC) curve. Classi…cation performance can be increased, however, if the information provided by multiple systems can be fused together to create a new, combined system. This research focuses on label-fusion as a common method to increase classi…cation performance and quantifying the bias that occurs when misspecifying the partitioning of the underlying event set. This partitioning will be de…ned in terms of what be called within and across label fusion. When incorrect assumptions are made about the partitioning of the event set, bias will occur and both the ROC curve and its optimal parameters will be incorrectly quanti…ed. In this work, we analyze the e¤ects of individual classi…cation system performance, correlation, and target environment on the magnitude of this performance bias. This work will then inspire the development of formulas to adjust optimal performance measures to appropriately re‡ect the fused system performance according to event set partitioning. As such, bias may be appropriately adjusted without redesigning the fused system, allowing greater use of currently fused systems across multiple platforms and environments.

Vector shoreline (VSL) data is potentially useful in ATR systems that distinguish between objects on land or water. Unfortunately available data such as the NOAA 1:250,000 World Vector Shoreline and NGA Prototype Global Shoreline data cannot be used by themselves to make a land/water determination because of the manner in which the data are compiled. We describe a data fusion approach for creating labeled VSL data using test points from Global 30 Arc-Second Elevation (GTOPO30) data to determine the direction of vector segments; i.e., whether they are in clockwise or counterclockwise order. We show consistently labeled VSL data be used to easily determine whether a point is on land or water using a vector cross product test.

The specification of geolocation accuracy requirements and their validation is essential for the proper performance of a Geolocation System and for trust in resultant three dimensional (3d) geolocations. This is also true for predicted accuracy requirements and their validation for a Geolocation System, which assumes that each geolocation produced (extracted) by the system is accompanied by an error covariance matrix that characterizes its specific predicted accuracy. The extracted geolocation and its error covariance matrix are standard outputs of (near) optimal estimators, either associated (internally) with the Geolocation System itself, or with a “downstream” application that inputs a subset of Geolocation System output, such as sensor data/metadata: for example, a set of images and corresponding metadata of the imaging sensor’s pose and its predicted accuracy. This output allows for subsequent (near) optimal extraction of geolocations and associated error covariance matrices based on the application’s measurements of pixel locations in the images corresponding to objects of interest. This paper presents recommended methods and detailed equations for the specification and validation of both accuracy and predicted accuracy requirements for a general Geolocation System. The specification/validation of accuracy requirements are independent from the specification/validation of predicted accuracy requirements. The methods presented in this paper are theoretically rigorous yet practical.

Algorithm selection is paramount in determining how to implement a process. When the results can be computed directly, an algorithm that reduces computational complexity is selected. When the results less binary there can be difficulty in choosing the proper implementation. Weighing the effect of different pieces of the algorithm on the final result can be difficult to find. In this research, we propose using a statistical analysis tool known as General Linear Hypothesis to find the effect of different pieces of an algorithm implementation on the end result. This will be done with transform based image fusion techniques. This study will weigh the effect of different transforms, fusion techniques, and evaluation metrics on the resulting images. We will find the best no-reference metric for image fusion algorithm selection and test this method on multiple types of image sets. This assessment will provide a valuable tool for algorithm selection to augment current techniques when results are not binary.

When geo-locating ground objects from a UAV, multiple views of the same object can lead to improved geo- location accuracy. Of equal importance to the location estimate, however, is the uncertainty estimate associated with that location. Standard methods for estimating uncertainty from multiple views generally assume that each view represents an independent measurement of the geo-location. Unfortunately, this assumption is often violated due to correlation between the location estimates. This correlation may occur due to the measurements coming from the same platform, meaning that the error in attitude or location may be correlated across time; or it may be due to external sources (such as GPS) having the same error in multiple aircraft. In either case, the geo-location estimates are not truly independent, leading to optimistic estimates of the geo-location uncertainty. For distributed data fusion applications, correlation-agnostic fusion methods have been developed that can fuse data together regardless of how much correlation may be present between the two estimates. While the results are generally not as impressive as when correlation is perfectly known and taken into account, the fused uncertainty results are guaranteed to be conservative and an improvement on operating without fusion. In this paper, we apply a selection of these correlation-agnostic fusion techniques to the multi-view geo-location problem and analyze their effects on geo-location and predicted uncertainty accuracy. We find that significant benefits can be found from applying these correlation agnostic fusion effects, but that they vary greatly in how well they estimate their own uncertainty.

Currently, when data is collected, it is usually collected for a specific need or situation. This includes text and image data. When a new need or situation arises, the data collection process repeats, often without referencing the original data collected for previous situations. Data Transpositioning is a search methodology that leverages the context of a previous manual search process, to formulate a new automated search with new results. As a result, the data collection process for one situation, can quickly be applied to another situation, but with less user effort. Thus, a set of new results can quickly be constructed without the user manually revisiting each of the originating sources. In the case of Content-Based Image Retrieval, the idea is to identify the content attributes of an image, such as a particular color, shape or texture, and apply changes to the originating query, and return a new set of results with the similar attributes. Data Transpositioning has been successfully applied to the result sets that contain text. Our goal is to continue this research beyond text to solve more complex problems in other domains, especially when image data are involved.

Feature extraction is a process used to reduce data dimensions using various transforms while preserving the discriminant characteristics of the original data. Feature extraction has been an important issue in pattern recognition since it can reduce the computational complexity and provide a simplified classifier. In particular, linear feature extraction has been widely used. This method applies a linear transform to the original data to reduce the data dimensions. The decision boundary feature extraction method (DBFE) retains only informative directions for discriminating among the classes. DBFE has been applied to various parametric and non-parametric classifiers, which include the Gaussian maximum likelihood classifier (GML), the k-nearest neighbor classifier, support vector machines (SVM) and neural networks. In this paper, we apply DBFE to deep neural networks. This algorithm is based on the nonparametric version of DBFE, which was developed for neural networks. Experimental results with the UCI database show improved classification accuracy with reduced dimensionality.

Networks of vision sensors are deployed in many settings, ranging from security needs to disaster response to environmental monitoring. Many of these setups have hundreds of cameras and tens of thousands of hours of video. The difficulty of analyzing such a massive volume of video data is apparent whenever there is an incident that requires foraging through vast video archives to identify events of interest. As a result, video summarization, that automatically extract a brief yet informative summary of these videos, has attracted intense attention in the recent years. Much progress has been made in developing a variety of ways to summarize a single video in form of a key sequence or video skim. However, generating a summary from a set of videos captured in a multi-camera network still remains as a novel and largely under-addressed problem. In this paper, with the aim of summarizing videos in a camera network, we introduce a novel representative selection approach via joint embedding and capped e2;1-norm minimization. The objective function is two-fold. The first is to capture the structural relationships of data points in a camera network via an embedding, which helps in characterizing the outliers and also in extracting a diverse set of representatives. The second is to use a capped e2;1-norm to model the sparsity and to suppress the influence of data outliers in representative selection. We propose to jointly optimize both of the objectives, such that embedding can not only characterize the structure, but also indicate the requirements of sparse representative selection. Extensive experiments on standard multi-camera datasets well demonstrate the efficacy of our method over state-of-the-art methods.

The Eigen-Template (ET) based closed-set feature extraction approach is extended to an open-set HRR-ATR framework to develop an Open Set Probabilistic Support Vector Machine (OSP-SVM) classifier. The proposed ET-OSP-SVM is shown to perform open set ATR on HRR data with 80% PCC for a 4-class MSTAR dataset.

Recent years have shown prolific advancements in smart infrastructures, allowing buildings of the modern world to interact with their occupants. One of the sought-after attributes of smart buildings is the ability to provide unobtrusive, indoor localization of occupants. The ability to locate occupants indoors can provide a broad range of benefits in areas such as security, emergency response, and resource management. Recent research has shown promising results in occupant building localization, although there is still significant room for improvement. This study presents a passive, small-scale localization system using accelerometers placed around the edges of a small area in an active building environment. The area is discretized into a grid of small squares, and vibration measurements are processed using a pattern matching approach that estimates the location of the source. Vibration measurements are produced with ball-drops, hammer-strikes, and footsteps as the sources of the floor excitation. The developed approach uses matched filters based on a reference data set, and the location is classified using a nearest-neighbor search. This approach detects the appropriate location of impact-like sources i.e. the ball-drops and hammer-strikes with a 100% accuracy. However, this accuracy reduces to 56% for footsteps, with the average localization results being within 0.6 m (α = 0.05) from the true source location. While requiring a reference data set can make this method difficult to implement on a large scale, it may be used to provide accurate localization abilities in areas where training data is readily obtainable. This exploratory work seeks to examine the feasibility of the matched filter and nearest neighbor search approach for footstep and event localization in a small, instrumented area within a multi-story building.

Structural health monitoring (SHM) is considered as an incentive multi-disciplinary technology for conditional assessment of infrastructure system. However, most classical output-only system identification methods based on a stationary assumption fail to achieve reliable results under non-stationary scenarios. Numerous techniques from the disciplines of multivariate statistics and pattern recognition in the field of structural damage detection have been developed, such as stochastic subspace identification or null-space and subspace damage identification. In order to obtain the accurate and quick structural damage assessment of a structure, measurement from all sensing nodes in the structure must be used. Therefore, dimension reduction and data visualization techniques for damage assessment need to be investigated. In this study a structural health monitoring method for damage identification and localization, which incorporated with the principal component analysis (PCA) based data compression and pattern recognition is developed. First, the frequency response function (FRF)-based damage assessment will be investigated. Based on the FRF to construct the Sammon map from all the measurements will be investigated. Sammon’s mapping is a nonlinear dimensional reduction algorithm which seeks to preserve distances as far as possible. Similarity among different data structure can be used to detect damage localization. Besides, the correlation of time-frequency analysis of data (Hilbert amplitude spectrum from WPT) among different measurement and the differences on the reconstructed data using multivariate AR-model are also used to detection damage. Verification of the proposed methods by using shaking table tests of two structures is demonstrated: One is focus on the damage of lower vibration modes and the other one is focus on the damage of high frequency modes.

Due to limited availability of DRC clean patterns during the process and RET recipe development, OPC recipes are not tested with high pattern coverage. Various kinds of pattern can help OPC engineer to detect sensitive patterns to lithographic effects. Random pattern generation is needed to secure robust OPC recipe. However, simple random patterns without considering real product layout style can’t cover patterning hotspot in production levels. It is not effective to use them for OPC optimization thus it is important to generate random patterns similar to real product patterns. This paper presents a strategy for generating random patterns based on design architecture information and preventing hotspot in early process development stage through a tool called Layout Schema Generator (LSG). Using LSG, we generate standard cell based on random patterns reflecting real design cell structure – fin pitch, gate pitch and cell height. The output standard cells from LSG are applied to an analysis methodology to assess their hotspot severity by assigning a score according to their optical image parameters - NILS, MEEF, %PV band and thus potential hotspots can be defined by determining their ranking. This flow is demonstrated on Samsung 7nm technology optimizing OPC recipe and early enough in the process avoiding using problematic patterns.

Topological pattern-based methods for analyzing IC physical design complexity and scoring resulting patterns to identify risky patterns have emerged as powerful tools for identifying important trends and comparing different designs. In this paper, previous work is extended to include analysis of layouts designed for the 7nm technology generation. A comparison of pattern complexity trends with respect to previous generations is made. In addition to identifying topological patterns that are unique to a particular design, novel techniques are proposed for scoring those patterns based on potential yield risk factors to find patterns that pose the highest risk.

Systematic yield detractors are normally expected to be identified by ATPG test result diagnostics. Different test patterns have been designed to test different functions. Test diagnostics can identify failed functions so that product engineers, based on testing results, can narrow down which block in the design performs this function. However, it is often hard to narrow down to a more specific region in a product. This paper will present a working flow for using design diffing techniques to extract layout structures and perform a geometry analysis flow combined with testing results to find most probable suspects that may cause noticeable yield loss.

At 28nm technology node and below, hot spot prediction and process window control across production wafers have become increasingly critical to prevent hotspots from becoming yield-limiting defects. We previously established proof of concept for a systematic approach to identify the most critical pattern locations, i.e. hotspots, in a reticle layout by computational lithography and combining process window characteristics of these patterns with across-wafer process variation data to predict where hotspots may become yield impacting defects [1,2]. The current paper establishes the impact of micro-topography on a 28nm metal layer, and its correlation with hotspot best focus variations across a production chip layout. Detailed topography measurements are obtained from an offline tool, and pattern-dependent best focus (BF) shifts are determined from litho simulations that include mask-3D effects. We also establish hotspot metrology and defect verification by SEM image contour extraction and contour analysis. This enables detection of catastrophic defects as well as quantitative characterization of pattern variability, i.e. local and global CD uniformity, across a wafer to establish hotspot defect and variability maps. Finally, we combine defect prediction and verification capabilities for process monitoring by on-product, guided hotspot metrology, i.e. with sampling locations being determined from the defect prediction model and achieved prediction accuracy (capture rate) around 75%

Engineers are continually faced with decisions about how much data they can collect. In this work, we present a statistically-based smart sampling methodology which can be used to target data collection and ensure that the risk to the product is clearly understood. Smart sampling combines knowledge of the distributions of control statistics with knowledge of the run length distributions they induce to balance the cost of information against the ability to respond to anomalies. We define and explore three characteristics that any sampling plan deemed “smart” must explicitly address: control errors that are associated with basing decisions on sample data, jeopardy that is associated with uncertainty about the true condition of the process, and the switching mechanism that controls the dynamic response to the latest information about the process. We show how the interplay of these characteristics can be exploited to comprehend the merits of a sampling plan. Practical examples of optical product inspection and process defectivity control are presented and explained.

This article is focused in developing an improved cluster ensemble method based cluster forests. Cluster forests (CF) is considered as a version of clustering inspired from Random Forests (RF) in the context of clustering for massive data. It aggregates intermediate Fuzzy C-Means (FCM) clustering results via spectral clustering since pseudo-clustering results are presented in the spectral space in order to classify these data sets in the multidimensional data space. One of the main advantages is the use of FCM, which allows building fuzzy membership to all partitions of the datasets due to the fuzzy logic whereas the classical algorithms as K-means permitted to build just hard partitions. In the first place, we ameliorate the CF clustering algorithm with the integration of fuzzy FCM and we compare it with other existing clustering methods. In the second place, we compare K-means and FCM clustering methods with the agglomerative hierarchical clustering (HAC) and other theory presented methods using data benchmarks from UCI repository.

In this paper, we treat the supervised data classification, while using the fuzzy random forests that combine the hardiness of the decision trees, the power of the random selection that increases the diversity of the trees in the forest as well as the flexibility of the fuzzy logic for noise. We will be interested in the construction of a forest of fuzzy decision trees. Our system is validated on nine standard classification benchmarks from UCI repository and have the specificity to control some data, to reduce the rate of mistakes and to put in evidence more of hardiness and more of interoperability.

In recent years, iris biometric systems have increased in popularity and have been proven that are capable of handling large-scale databases. The main advantage of these systems is accuracy and reliability. A proper iris patterns classification is expected to reduce the matching time in huge databases. This paper presents an iris indexing technique based on Local Intensity Order Pattern. The performance of the present approach is evaluated on UPOL database and is compared with other recent systems designed for iris indexing. The results illustrate the potential of the proposed method for large scale iris identification.