Text mining methods can find large scale trends within research communities. For example, using stable Latent Dirichlet Allocation (a topic modeling algorithm) this study found 10 major topics in 35,391 SE research papers from 34 leading SE venues over the last 25 years (divided, evenly, between conferences and journals). Out study also shows how those topics have changed over recent years. Also, we note that (in the historical record) mono-focusing on a single topic can lead to fewer citations than otherwise. Further, while we find no overall gender bias in SE authorship, we note that women are under-represented in the top-most cited papers in our field. Lastly, we show a previously unreported dichotomy between software conferences and journals (so research topics that succeed at conferences might not succeed at journals, and vice versa). An important aspect of this work is that it is automatic and quickly repeatable (unlike prior SE bibliometric studies that used tediously slow and labor intensive methods). Automation is important since, like any data mining study, its conclusions are skewed by the data used in the analysis. The automatic methods of this paper make it far easier for other researchers to re-apply the analysis to new data, or if they want to use different modeling assumptions.

As code search permeates most activities in software development,code-to-code search has emerged to support using code as a query and retrieving similar code in the search results. Applications include duplicate code detection for refactoring, patch identification for program repair, and language translation. Existing code-to-code search tools rely on static similarity approaches such as the comparison of tokens and abstract syntax trees (AST) to approximate dynamic behavior, leading to low precision. Most tools do not support cross-language code-to-code search, and those that do, rely on machine learning models that require labeled training data. We present Code-to-Code Search Across Languages (COSAL), a cross-language technique that uses both static and dynamic analyses to identify similar code and does not require a machine learning model. Code snippets are ranked using non-dominated sorting based on code token similarity, structural similarity, and behavioral similarity. We empirically evaluate COSAL on two datasets of 43,146Java and Python files and 55,499 Java files and find that 1) code search based on non-dominated ranking of static and dynamic similarity measures is more effective compared to single or weighted measures; and 2) COSAL has better precision and recall compared to state-of-the-art within-language and cross-language code-to-code search tools. We explore the potential for using COSAL on large open-source repositories and discuss scalability to more languages and similarity metrics, providing a gateway for practical,multi-language code-to-code search.

Successful cross-language clone detection could enable researchers and developers to create robust language migration tools, facilitate learning additional programming languages once one is mastered, and promote reuse of code snippets over a broader codebase. However, identifying cross-language clones presents special challenges to the clone detection problem. A lack of common underlying representation between arbitrary languages means detecting clones requires one of the following solutions: 1) a static analysis framework replicated across each targeted language with annotations matching language features across all languages, or 2) a dynamic analysis framework that detects clones based on runtime behavior.

Video and multi-media sharing is a significant activity on social media platforms. Learning patterns of activities using raw video data is computationally intensive and impractical, and manual inspection is not scalable and prohibitively expensive. An alternate strategy is to learn information about video content using far less compute intensive metadata surrogates. This paper describes a video recommender tool implemented in GovCloud using a novel approach of using lightweight video metadata to learn and classify video content. In contrast to popular video recommender systems that use consumption models for classification, the new approach used in our tool is based solely on the video metadata along with domain expertise used to truth a relatively small subset of relevant video content. The tool is very user-friendly and captures practical knowledge of the user resulting in good learning model. The architecture and implementation specifics of the tool is outlined in this paper. The classifier performance using metadata from tens of thousands of real postings exceeds 90% for both recall and ROC metrics. This tool has shown promise in providing a console for aggregating social media videos for analysts to train the system consistent with the context and task at hand.

This paper introduces Data-Driven Search-based Software Engineering (DSE), which combines insights from Mining Software Repositories (MSR) and Search-based Software Engineering (SBSE). While MSR formulates software engineering problems as data mining problems, SBSE reformulate Software Engineering (SE) problems as optimization problems and use meta-heuristic algorithms to solve them. Both MSR and SBSE share the common goal of providing insights to improve software engineering. The algorithms used in these two areas also have intrinsic relationships. We, therefore, argue that combining these two fields is useful for situations (a) which require learning from a large data source or (b) when optimizers need to know the lay of the land to find better solutions, faster.

The processing demands on video analytics calls for special design considerations to achieve scalability. Numerous factors influence the running time of an analytics job. The time consumed for raw computing can be improved by well-engineered approaches to execute certain sub-tasks. High scalability can be achieved by selectively distributing computational components. We elucidate such factors that aid scalability and present design choices for architecting them. The principles outlined in this research were used to implement a distributed on-demand video analytics system that was prototyped for the use of forensics investigators in law enforcement. The system was tested in the wild using video files as well as a commercial Video Management System supporting more than 100 surveillance cameras as video sources. The architectural considerations of this system are presented. Issues to be reckoned with in implementing a scalable distributed on-demand video analytics system are highlighted.

When Requirements Engineering(RE) models are unreasonably complex, they cannot support efficient decision making. SHORT is a tool to simplify that reasoning by exploiting the "key" decisions within RE models. These "keys" have the property that once values are assigned to them, it is very fast to reason over the remaining decisions. Using these "keys", reasoning about RE models can be greatly SHORTened by focusing stakeholder discussion on just these key decisions.This paper evaluates the SHORT tool on eight complex RE models. We find that the number of keys are typically only 12% of all decisions. Since they are so few in number, keys can be used to reason faster about models. For example, using keys, we can optimize over those models (to achieve the most goals at least cost) two to three orders of magnitude faster than standard methods. Better yet, finding those keys is not difficult: SHORT runs in low order polynomial time and terminates in a few minutes for the largest models.

The use of array readers (i.e., multiple read elements positioned to read stored data on multiple adjacent tracks) is one of the recent concepts to improve the areal density of magnetic recording systems. Array readers provide diversity gain to deal with noise and enable the handling of inter-track interference (ITI) as well as multi-track detection. In this paper, an array-reader two-track (AR2T) detection system is studied. We present the design of 2-D equalizer that jointly processes array readback signal streams over two tracks. Such a 2-D equalizer can be designed to approximate a 1-D or a 2-D partial-response (PR) target. It is shown that 2-D PR target can result in less residual ITI. For the detection of signals corresponding to 2-D PR target, we propose a novel symbol-based detection algorithm that processes two input signal streams jointly. We also implement a pattern-dependent noise-predictive (PDNP) version of proposed detector. Simulation results show that more than 4 dB performance gain can be achieved by the AR2T detection system compared to the traditional single-reader single-track detection system under aggressive track density conditions.

Building prediction models for suggestive knowledge from multiple sources dynamically is of great interest from a clinical decision support point of view. This is valuable in situations where the local clinical data repository does not have sufficient number of records to draw conclusions from. However, due to privacy concerns, hospitals are reluctant to divulge patient records. Consequently, a distributed model building mechanism that can use just the statistics from multiple hospitals' databases is valuable. Our DIDT algorithm builds a model in that fashion. In this study, using National Inpatient Sample (NIS) data for 2009, we demonstrate that DIDT algorithm can be used to help collaboratively build a better decision-making model in situations where hospitals have small number of records that are insufficient to make good local models. Based on 262 attributes used for model building, we showed that 9 collaborating hospitals each with less than 100 cases of hospitalizations related to diabetes were able to achieve 9.9% improvement in accuracies of hospitalization prediction collectively using a distributed model as compared to relying on local models developed on their own. When relying on local risk prediction models for diabetes at these 9 hospitals, 159 of 357 patients were misclassified and prediction was impossible for another 16 patients. Our integrated model reduced the misclassification to 138 effectively providing accurate early diagnostics to 37 additional patients. We also introduce the concept of banding to improve DIDT algorithm so as to logically combine multiple hospitals when large number of hospitals is involved for reduction in cross-validation folds.

We present a multiscale adaptive search algorithm for efficiently searching an unknown number of stationary targets using a team of multiple mobile sensors. We first derive a Spectral Multiscale Coverage (SMC) control law for a Dubins vehicle model. Given a search prior, the SMC control leads to uniform coverage dynamics for the mobile sensors such that the amount of time spent observing a region is proportional to finding a target in it. In order to make the search robust to sensor uncertainties and Automatic Target Detection algorithm errors (i.e. false alarm, missed detections), we combine the SMC control with decision and estimation theoretic techniques. As new targets are discovered we use the Sequential Ratio Probability Test and Recursive Least Squares estimation to quantify the current uncertainty in target detection and location, respectively. This uncertainty is used to update the search prior so as to balance exploitation (reduce uncertainty in state of already discovered potential targets) and exploration (discover new targets). We demonstrate this adaptive search methodology in a high fidelity simulation environment and show an improved performance over lawnmower type search.

OBJECTIVE: To evaluate the benefit of a health information exchange (HIE) between hospitals, we examine the rate of crossover among neurosurgical inpatients treated at Emory University Hospital (EUH) and Grady Memorial Hospital (GMH) in Atlanta, Georgia. To inform decisions regarding investment in HIE, we develop a methodology analyzing crossover behavior for application to larger more general patient populations. DESIGN: Using neurosurgery inpatient visit data from EUH and GMH, unique patients who visited both hospitals were identified through classification by name and age at time of visit. The frequency of flow patterns, including time between visits, and the statistical significance of crossover rates for patients with particular diagnoses were determined. MEASUREMENTS: The time between visits, flow patterns, and proportion of patients exhibiting crossover behavior were calculated for the total population studied as well as subpopulations. RESULTS: 5.25% of patients having multiple visits over the study period visited the neurosurgical departments at both hospitals. 77% of crossover patients visited the level 1 trauma center (GMH) before visiting EUH. LIMITATIONS: The true patient crossover may be under-estimated because the study population only consists of neurosurgical inpatients at EUH and GMH. CONCLUSION: We demonstrate that detailed analysis of crossover behavior provides a deeper understanding of the potential value of HIE.

We present a continuous-space multiscale adaptive search (MAS) algorithm for single or multiple searchers that finds a stationary target in the presence of uncertainty in sensor diameter. The considered uncertainty simulates the influence of the changing environment and terrain as well as adversarial actions that can occur in practical applications. When available, information about the foliage areas and a priori distribution of the target position is included in the MAS algorithm. By adapting to various uncertainties, MAS algorithm reduces the median search time to find the target with a probability of detection of at least PD and a probability of false alarm of at most PFA. We prove that MAS algorithm discovers the target with the desired performance bounds PD and PFA. The unique features of the MAS algorithm are realistic second-order dynamics of the mobile sensors that guarantees uniform coverage of the surveyed area and a two-step Neyman-Pearson-based decision-making process. Computer simulations show that MAS algorithm performs significantly better than lawnmower-type search and billiard-type random search. Our tests suggest that the median search time in the MAS algorithm may be inversely proportional to the number of participating searchers. As opposed to lawnmower search, the median search time in the MAS algorithm depends only logarithmically on the magnitude of uncertainty.

Collaborators use vocabulary germane to the domain in context. Collaboration applications and collaborating systems use vocabularies at different (lower) layers that are specific to the state in which they execute. Distributed, yet collaborative domain-specific applications have demonstrated success

Due to the increase in federated nature of collaboration applications, users from multiple institutions have the potential to participate in activities centered around common regions of interest. However, existing technologies address external users at a coarse grained level. Consequently, mechanism

Maximum likelihood (ML) carrier-frequency offset (CFO) estimation for orthogonal frequency-division multiple-access (OFDMA) uplink with generalized carrier-assignment scheme (GCAS) is a complex multi-parameter estimation problem. The computational complexity of ML solution based on a multi-dimensional exhaustive search is prohibitive. The existing sub-optimal solutions reduce the complexity by replacing the multi-dimensional search with a sequence of single-dimensional searches. However, these solutions suffer from either poor estimation accuracy or being still of fairly high complexity. In this paper, we propose a new approach called divide-and-update frequency estimator (DUFE), for CFO estimation. Compared with the existing approaches, the proposed DUFE has lower computational complexity while maintaining high estimation accuracy similar to that of the exact ML solution. Performance and complexity comparisons are provided, along with numerical results to illustrate the effectiveness of the proposed method.

Maximum likelihood (ML) carrier-frequency offset (CFO) estimation in the uplink of an orthogonal frequency-division multiple-access (OFDMA) system with a generalized carrier- assignment scheme (GCAS) is a complex multiple-parameter estimation problem. The computational complexity of the ML solution based on a multi-dimensional exhaustive search is prohibitive. The existing solutions reduce the complexity by replacing the multidimensional search with a sequence of single-dimensional searches. However, those solutions suffer from either poor estimation accuracy or is still of high complexity. In this paper, we propose a new approach called as the divide-and-update frequency estimator (DUFE), which outperforms the existing solutions in the sense that it has lower computational complexity while maintaining high estimation accuracy similar to the exact ML solution. Performance and complexity comparisons are provided with numerical results to illustrate the effectiveness of the proposed method.

Application of existing maximum likelihood channel estimation (MLE) in orthogonal frequency division multiplexing (OFDM) systems requires knowledge of the effective length of channel impulse response (ELCIR) for achieving optimum performance. The analysis shows that the mean-squared error (MSE) is linearly related to ELCIR. Tracking the variation in ELCIR is thus very important for conventional MLE. But, incorporating a run-time update of ELCIR into the ML estimator turns out to be computationally expensive. Therefore, a modified ML channel estimator, which combines the ML estimation with a frequency-domain smoothing technique, is proposed. The proposed method introduces no extra complexity, and its performance has been proved using theoretical analysis and simulations to be robust to variation in ELCIR. Numerical results are provided to show the effectiveness of the proposed estimator under time-invariant and time-variant channel conditions.

In this paper, we investigate various frequency-domain channel estimators for orthogonal frequency division multiplexing (OFDM) systems with virtual carriers (VCs). As conventional estimators cannot estimate the channel transfer function (CTF) at VCs, DFT-based estimators are not directly applicable in OFDM systems with VCs. To circumvent this, we present least squares (LS), generalized linear minimum mean square error (GLMMSE) and generalized singular value decomposition (GSVD) methods to estimate the CTF at VCs. Further, by exploiting the received data samples at VCs, we describe a novel noise variance estimation method. We also analyze the performance degradation due to estimated noise variance and the channel correlation mismatch. Performance comparison shows that the discrete Fourier transform (DFT)-based channel estimators incorporating the proposed CTF estimation at VCs achieve at least 6 dB gain when MMSE= 10-3 compared to frequency-domain LS estimator. Further, even though the DFT-based LS estimator with VCs performs worst among all the DFT-based estimators, it is the most robust estimator.

We present an analytical solution for optimum partial response (PR) in Viterbi-detector-based communication receivers by maximizing the detection signal-to-noise ratio

Timing recovery is essential to signal detection in magnetic recording. The paper presents the development of timing acquisition techniques for partial response (PR) equalized perpendicular magnetic recording channels. Because preamble sequences are used to help in the acquisition process, the optimality of different preamble sequences is investigated for different user densities. Variable threshold detectors (VTDs), designed to tackle hang up and false lock problems, are presented. A new timing error detector (TED), which is very. simple to implement, is developed by exploiting the features of the preamble sequence and PR target. Using analytical and simulation studies, the new TED is shown to be optimal for the chosen target and preamble.

Runlength limited (RLL) codes are essential in recording systems for minimizing distortions and maintaining bit synchronization. In this paper, we investigate the application of turbo codes on RLL (1,7) coded partial response equalized recording channels. We proposed a 'combined trellis' approach for doing 'soft-in soft-out' (SISO) channel detection and RLL decoding jointly. This approach makes the implementation of turbo equalization easier since it eliminates the need for a 'SISO RLL encoder' in the feedback path from turbo decoder to channel detector. Simulation studies on magnetic recording channels show that our approach provides a coding gain of about 4 dB compared to Viterbi detection at 10/sup -6/ bit error rate.

Timing-recovery and gain-control play important role in communication systems. In this paper, we propose reliable acquisition, tracking and gain control algorithms for fixed delay tree search with decision feedback (FDTS/DF) detector on magnetic recording channel. First, we propose a modified Mueller and Muller algorithm for timing recovery during tracking period, which is based on selective transitions in the samples at the detector input. Next, we give the modified threshold detection scheme that quickly converges to the correct sampling phase without hang-up problems during acquisition period. Bit-by-bit simulations are included first to show the reliability in tracking performance of the timing-recovery and gain-control loops, and then to show the robustness and accuracy of the fast acquisition algorithm.

The application of iterative detection using turbo codes for signal detection in magnetic recording channels has become a topic of intense research. Interesting results have been reported for partial response equalized recording channels. This paper investigates the performance of turbo codes on decision feedback equalized (DFE) recording channels. The channel detector used is a soft-in soft-out parallel DFE (PDFE) detector. Simulation results shows a coding gain of more than 3 dB at a bit error rate of 10/sup -6/. Further, the error propagation in PDFE has been significantly minimized by the de-interleaver between PDFE and turbo decoder, and it does not influence the overall burst error performance.

The study of error-burst statistics is important for all detection systems, and more so for the decision feedback class. In data storage applications, many detection systems use decision feedback in one form or another. Fixed-delay tree search with decision feedback (FDTS/DF) and decision feedback equalization (DFE) are the direct forms, whereas the partial response detectors such as the reduced state sequence estimator (RSSE) and noise predictive maximum likelihood (NPML) detectors are the other forms. Although DF reduces the system complexity, it is inevitably linked with error propagation (EP), which can be quantified using error-burst statistics. Analytical evaluation of these statistics is difficult, if not impossible, because of the complexity of the problem. Hence, the usual practice is to use computer simulations. However, the computational time in traditional bit-by-bit simulations can be prohibitive at meaningful signal-to-noise ratios. In this paper, we propose a novel approach for fast estimation of error-burst statistics in FDTS/DF detectors, which is also applicable to other detection systems. In this approach, error events are initiated more frequently than natural by artificially injecting noise samples. These noise samples are generated using a transformation that results in significant reduction in computational complexity. Simulation studies show that the EP performance obtained by the proposed method matches closely with those obtained by bit-by-bit simulations, while saving as much as 99% of simulation time.

In this paper, we investigate the effect of timing offset on the performance of MDFE (multi-level decision feedback equalization) and PR-VD (partial response with Viterbi detection) detectors, in magnetic recording applications. We develop analytical approaches to evaluate the performance using error event rate as a measure. The main step in this evaluation is the computation of the probability density function of residual ISI (intersymbol interference) arising from mis-equalization and timing offset. The resulting performance curves are compared with those obtained from simulations. The detection performance due to steady-state phase jitter is also evaluated. The results show that MDFE is better than PR-VD in terms of timing sensitivity.

In this letter, we propose a method for blind separation of d co-channel BPSK signals arriving at an antenna array. Our method involves two steps. In the first step, the received data vectors at the output of the array is grouped into 2/sup d/ clusters. In the second step, we assign the 2/sup d/ d-tuples with /spl plusmn/1 elements to these clusters in a consistent fashion. From the knowledge of the cluster to which a data vector belongs, we estimate the bits transmitted at that instant. Computer simulations are used to study the performance of our method.>

An adaptive approach for estimating all (or some) of the orthogonal eigenvectors of the data covariance matrix (of a time series consisting of real narrowband signals in additive white noise) is presented. The inflation approach is used to estimate each of these vectors as minimum eigenvectors (eigenvectors corresponding to the minimum eigenvalue) of appropriately constructed symmetric positive definite matrices. This reformulation of the problem is made possible by the fact that the problem of estimating the minimum eigenvector of a symmetric positive definite matrix can be restated as the unconstrained minimization of an appropriately constructed nonlinear nonconvex cost function. The modular nature of the algorithm that results from this reformation makes the proposed approach highly parallel, resulting in a high-speed adaptive approach for subspace estimation.>