Time Series Data Streams Research Articles

Deep neural networks to enable real-time multimessenger astrophysics

Gravitational wave astronomy has set in motion a scientific revolution. To further enhance the science reach of this emergent field, there is a pressing need to increase the depth and speed of the gravitational wave algorithms that have enabled these groundbreaking discoveries. To contribute to this effort, we introduce Deep Filtering, a new highly scalable method for end-to-end time-series signal processing, based on a system of two deep convolutional neural networks, which we designed for classification and regression to rapidly detect and estimate parameters of signals in highly noisy time-series data streams. We demonstrate a novel training scheme with gradually increasing noise levels, and a transfer learning procedure between the two networks. We showcase the application of this method for the detection and parameter estimation of gravitational waves from binary black hole mergers. Our results indicate that Deep Filtering significantly outperforms conventional machine learning techniques, achieves similar performance compared to matched-filtering while being several orders of magnitude faster thus allowing real-time processing of raw big data with minimal resources. More importantly, Deep Filtering extends the range of gravitational wave signals that can be detected with ground-based gravitational wave detectors. This framework leverages recent advances in artificial intelligence algorithms and emerging hardware architectures, such as deep-learning-optimized GPUs, to facilitate real-time searches of gravitational wave sources and their electromagnetic and astro-particle counterparts.

A Novel Multi-resolution Representation for Streaming Time Series

Along with the coming of IoT (Internet of things) era, massive numbers of instruments and applications in various fields are continuously producing oceans of time series stream data, which could be characterized by its large amount, high dimensionality and continuity nature. In order to carry out different kinds of data mining tasks (similarity search, classification, clustering, prediction etc.) based on streaming time series efficiently and effectively, segmentation and representation which segment a streaming time series into several subsequences and provide more compact representation for the raw data, should be done as the first step. With the virtue of solid theoretical foundations, piecewise linear representation (PLR) has been gained success in yielding more compact representation and fewer segments, however, the current state of art PLR methods have their own flaws. In this paper, we propose a novel online time series segmentation algorithm called continuous segmentation and multi-resolution representation algorithm based on turning points (CSMR_TP), which partitions the streaming time series by a set of temporal feature points and represents the time series flexibly. Our method can not only generate more accurate approximation than the state-of-the-art of PLR algorithm, but also represent the streaming time series in a more flexible way to meet the diverse needs of users. Extensive experiments on different kinds of typical time series datasets have been conducted to demonstrate the superiorities of our method.

Low Flow Trends and Frequency Analysis in the Blue Nile Basin, Ethiopia

Low flow analysis provides crucial information for the planning and design water resource development, risk assessment and environmental flow management. Understanding the low flow regimes and evaluating the magnitudes for incorporating in water resources management is vital for the countries like Ethiopia where demand for water is increasing. However, there were hardly enough studies in understanding the trends of low flow and frequency analysis. Therefore, this study focuses on evaluation of the trends in low flows and regional low flow analysis in the Blue Nile Basin, Ethiopia. In order to carry out the study, 15 river sub-basins in the Blue Nile Basin were selected based on the long term data availability and presence of quality of data. The 3-day sustained low flow (3d-slf), the 7-day sustained low flow (7d-slf) and the 14-day sustained low flow (14d-slf) models were used to extract the data from the daily time series stream data obtained from MoWIE. Trends in low flow were analyzed separately by using Mann-Kendall (MK) trend test. Low flow frequency analysis was used to estimate the long term low flow quantiles. In addition, regional analysis for estimating the quantiles for ungaged catchments was also developed based on the regional growth curve and catchment characteristic of drainage basins. The results indicated that 3d-slf, 7d-slf and 14d-slf models of low flow series indicated no significant difference for each station at 95% CI. Out of the 15 selected stations, 12 of stations have indicated decreasing; two stations indicated increasing and remaining one station with no trend. Mainly decreasing trend was associated with the land cover and climate change which results in increasing runoff and evapotranspiration respectively. Weibull distribution—GEV and LGN was found best fit based on the L-Moment Ratio Diagram (L-MRD). Hence quantile estimations have indicated diminishing magnitudes of low flow quintiles for 2 - 500 years return periods. Regional low frequency analysis has provided a very good relationship between discharge and catchment characteristics with an R2 of 0.72. Where area (A) and rainfall (R) followed by slope were found sensitive to compute in developing the regional region equations between mean low flows and the physiographic data. This study indicated that there needs to be a new water management scenario and adaptation mechanism of climate change and land use land cover dynamics for utilizing water resource in the Blue Nile Basin.

Feature Extraction from Sensor Data Streams for Optimizing Grinding Condition

A visualization method for time-series sensing data was designed to optimize grinding condition. The fluctuation pattern of time-series data streams can be visualized as a white and black pattern by utilizing the spindle power change rate average. The designed visualization method was applied to a condition monitoring in lapping operation. The relation between the fallout abrasive grain content and lapping behaviour was experimentally examined. In the lapping with grinding fluid containing no fallout abrasive, the spindle power decreased in a monotone manner with lapping time, while in the lapping with fallout abrasive, the spindle power decreased with lapping time up to 20s of lapping and then tended to converge on a constant value. The spindle power change rate average displayed as a white and black pattern reproduced the changes of spindle power very well. The appearance probability of white or black pattern has a strong relation with the fallout abrasive content and the designed data processing scheme could make possible to predict the grinding fluid condition from the easy-handling grinding test.

Recovery and self-efficacy of a newly trained certified peer specialist following supplemental weekly group supervision: A case-based time-series analysis

ABSTRACTThis single-case time-series study explored the benefits of adding a supervision/support group for peer specialists. Six certified peer specialists, employed by a small community-based peer support program, provided weekly recordings of mental health recovery (as measured by the Recovery Assessment Scale) and peer specialist self-efficacy (measured with an inventory developed by the authors) during an 8-week baseline phase and a 28-week intervention (weekly supervision group) phase. Data from the peer specialist with the most consistent supervision group attendance and greatest number of weekly recordings were chosen for intensive analysis using Simulation Modeling Analysis for Brief Time-Series Data Streams. Results indicated that the peer specialist experienced significant increases in recovery and work-related self-efficacy during the group supervision phase. There was also a statistically significant association between her improvements in mental health recovery and her increases in peer speci...

An Efficient Time Series Subsequence Pattern Mining and Prediction Framework with an Application to Respiratory Motion Prediction

Traditional time series analysis methods are limited on some complex real-world time series data. Respiratory motion prediction is one of such challenging problems. The memory-based nearest neighbor approaches haveshown potentials in predicting complex nonlinear time series compared to many traditional parametric prediction models. However, the massive time series subsequences representation, the similarity distance measures, the number of nearest neighbors, and the ensemble functions create challenges as well as limit the performance of nearest neighbor approaches in complex time series prediction. To address these problems, we propose a flexible time series pattern representation and selection framework, called the orthogonalpolynomial-based variant-nearest-neighbor (OPVNN) approach. For the respiratory motion prediction problem, the proposed approach achieved the highest and most robust prediction performance compared to the state-of-the-art time series prediction methods. With a solid mathematical and theoretical foundation in orthogonal polynomials, the proposed time series representation, subsequence pattern mining and prediction framework has a great potential to benefit those industry and medical applications that need to handle highly nonlinear and complex time series data streams, such as quasi-periodic ones.

Discovering sub-patterns from time series using a normalized cross-match algorithm

Time series data stream mining has attracted considerable research interest in recent years. Pattern discovery is a challenging problem in time series data stream mining. Because the data update continuously and the sampling rates may be different, dynamic time warping (DTW)-based approaches are used to solve the pattern discovery problem in time series data streams. However, the naive form of the DTW-based approach is computationally expensive. Therefore, Toyoda proposed the CrossMatch (CM) approach to discover the patterns between two time series data streams (sequences), which requires only O(n) time per data update, where n is the length of one sequence. CM, however, does not support normalization, which is required for some kinds of sequences (e.g. stock prices, ECG data). Therefore, we propose a normalized-CrossMatch approach that extends CM to enforce normalization while maintaining the same performance capabilities.

Comparing and Combining Time Series Trajectories Using Dynamic Time Warping

This research proposes the application of dynamic time warping (DTW) algorithm to analyse multivariate data from virtual reality training simulators, to assess the skill level of trainees. We present results of DTW algorithm applied to trajectory data from a virtual reality haptic training simulator for epidural needle insertion. The proposed application of DTW algorithm serves two purposes, to enable (i) two trajectories to be compared as a similarity measure and also enables (ii) two or more trajectories to be combined together to produce a typical or representative average trajectory using a novel hierarchical DTW process.Our experiments included 100 expert and 100 novice simulator recordings. The data consists of multivariate time series data-streams including multi-dimensional trajectories combined with force and pressure measurements.Our results show that our proposed application of DTW provides a useful time-independent method for (i) comparing two trajectories by providing a similarity measure and (ii) combining two or more trajectories into one, showing higher performance compared to conventional methods such as linear mean. These results demonstrate that DTW can be useful within virtual reality training simulators to provide a component in an automated scoring and assessment feedback system.

NSPRING: the SPRING extension for subsequence matching of time series supporting normalization

Mining sequences and patterns in time series data streams is fast becoming a common practice in today's world. The rapid progress of data collection and web technologies yields tremendous growth of flowing data in various complex forms that need to be analyzed in real time. Traditional data mining methods that typically require the process data to be scanned repeatedly are not feasible for stream data applications. However, new techniques like SPRING attempt to address these challenges by identifying sequences of patterns on time series streams, thus reducing the complexity to be linear in both time and space. Unfortunately, SPRING does not support data normalization, which renders it to be not applicable for most data sets. In this paper, we are proposing an approach called NSPRING based on SPRING that extends the advantages of SPRING, e.g., low in time and space complexity, while it can support normalization. Furthermore, NSPRING retains similar mining accuracy to SPRING.

A Survey on Effective Pattern Matching in Uncertain Time Series Stream Data

A Survey on Effective Pattern Matching in Uncertain Time Series Stream Data

A Pattern Query Strategy Based on Semi-supervised Machine Learning in Distributed WSNs

Hardware of Distributed WSNs is limited, which turns into a big challenge. In recent years, there are some pattern-based query strategies have been proposed. But compared to the methods based on machine learning, these methods need high energy consumption in the calculation of distance. In this paper, we propose a pattern-based query strategy based on semi-supervised machine learning in wireless sensor network. The strategy uses historical time series data streams for o†ine machine learning. First of all, we flnd out the flxed length subsequences and screen out the suitable data sequences as training dataset and sample dataset. And then, because labeled data is less in practical application, we do label propagation with graph-based semi-supervised machine learning algorithm, Local and Global Consistency (LGC), to increase the number of labeled data sequences. Next, we use the supervised learning algorithm Support Vector Machine (SVM) to train a classifler. Finally, for a new data sequence, we can directly get its category through the classifler. Experimental results show that the performance of our strategy is almost better than other pattern-based query strategies on classiflcation accuracy and energy consumption.

Analyzing Well Events To Increase Oil Recovery

Technology Update The human mind, presented with a series of time data, will generally look for common features, label them as events, and seek correlations against other timelines. This approach is fallible because questionable data might be selected in attempting to find supporting evidence for a perceived pattern. A better alternative would be to establish an automated, comprehensive, dispassionate, independent, and statistically valid process for pattern matching. It is a major challenge to develop such a process. But if all the objectives are met, then many evaluations can be carried out using an “event-based” analysis. Instrumented systems and the analysis of their time-series data can provide a range of events and channels through which those events can interact. Some influences will be automatically transmitted by closed-loop process control schemes. Others will be prescribed by written operating practices, and still others will be the result of “custom and practice” on the part of operators, supervisors, and planners. Despite the presence of these systematic, nonreservoir-derived reasons for patterns of change, the historical daily production data is the most direct measurement of the hydrocarbon fluids being extracted from the reservoir. The business case to pursue increases in efficiency of extraction and total recovery, while simultaneously being able to accurately assess the effectiveness of measures taken to support, sustain, and maximize that production, is clear. New ‘Top-Down’ Analysis Understanding the flow from each well as a result of an agreed allocation process, based on the assessment of flow data from the well’s tests, allows history matching. Typically, one or more reservoir simulation models are tuned to the observed flow and pressure data. These processes represent a “bottom-up” approach to understanding reservoir performance, through which a development plan can be monitored and updated as necessary. However, BP has been working on a separate but entirely complementary “top-down” approach in which operational data is analyzed without preconceptions about the reservoir structure. This approach allows the rapid formulation of a workable “model,” thus avoiding the need for an involved history-matching procedure. The top-down approach uses the capacitance resistivity model (CRM) combined with an analysis of the operational time-series data based on “events” in that data. Event-Based Approach Firstly, one must choose the production and injection wells to include for the asset being studied. These will be sufficiently instrumented to provide a time-series data stream containing events, which are distinguishable from the unavoidable sources of measurement noise. The wells will be material to the oil recovery plan and will include some key injection wells, in addition to producers in which output might be enhanced by the injectors. Next, one must choose a period for analysis. Recent data is likely to be the most instructive in terms of potentially assessing probable future events, but better data-quality periods may exist. Quality may be measured in terms of working instrumentation, constant or stable numbers of injection and production wells in service, and the absence of changes in operating regime, such as artificial lift or the breakthrough of water or gas into production wells. Then, one must choose the well attribute that will be used as the variable to indicate the occurrence of an event. Typically, the choice will be an allocated (or possibly measured) flow, an aspect of production, such as water cut or gas/oil ratio, or a direct intensive measurement, such as a pressure or a temperature. Similarly, injection wells will have some (typically) corresponding attribute, such as injection rate or pressure. The event-based analysis begins with the marking of events for each production and injection well. Parameters are used to control the relative occurrence of events. Some iteration is typically required until a set of events that make sense organically and pass visual assessments has been achieved. Visualizations made using the association software assist with events evaluation.

Maximum error-bounded Piecewise Linear Representation for online stream approximation

Given a time series data stream, the generation of error-bounded Piecewise Linear Representation (error-bounded PLR) is to construct a number of consecutive line segments to approximate the stream, such that the approximation error does not exceed a prescribed error bound. In this work, we consider the error bound in $$L_\infty $$ L ? norm as approximation criterion, which constrains the approximation error on each corresponding data point, and aim on designing algorithms to generate the minimal number of segments. In the literature, the optimal approximation algorithms are effectively designed based on transformed space other than time-value space, while desirable optimal solutions based on original time domain (i.e., time-value space) are still lacked. In this article, we proposed two linear-time algorithms to construct error-bounded PLR for data stream based on time domain, which are named OptimalPLR and GreedyPLR, respectively. The OptimalPLR is an optimal algorithm that generates minimal number of line segments for the stream approximation, and the GreedyPLR is an alternative solution for the requirements of high efficiency and resource-constrained environment. In order to evaluate the superiority of OptimalPLR, we theoretically analyzed and compared OptimalPLR with the state-of-art optimal solution in transformed space, which also achieves linear complexity. We successfully proved the theoretical equivalence between time-value space and such transformed space, and also discovered the superiority of OptimalPLR on processing efficiency in practice. The extensive results of empirical evaluation support and demonstrate the effectiveness and efficiency of our proposed algorithms.

Simulation modelling analysis for small sets of single-subject data collected over time

The behavioural data yielded by single subjects in naturalistic and controlled settings likely contain valuable information to scientists and practitioners alike. Although some of the properties unique to this data complicate statistical analysis, progress has been made in developing specialised techniques for rigorous data evaluation. There are no perfect tests currently available to analyse short autocorrelated data streams, but there are some promising approaches that warrant further development. Although many approaches have been proposed, and some appear better than others, they all have some limitations. When data sets are large enough (∼30 data points per phase), the researcher has a reasonably rich pallet of statistical tools from which to choose. However, when the data set is sparse, the analytical options dwindle. Simulation modelling analysis (SMA; described in this article) is a relatively new technique that appears to offer acceptable Type-I and Type-II error rate control with short streams of autocorrelated data. However, at this point, it is probably too early to endorse any specific statistical approaches for short, autocorrelated time-series data streams. While SMA shows promise, more work is needed to verify that it is capable of reliable Type-I and Type-II error performance with short serially dependent streams of data.

Sequential Pattern Mining for Uncertain Data Streams using Sequential Sketch

Uncertainty is inherent in data streams, and present new challenges to data streams mining. For continuous arriving and large size of data streams, modeling sequences of uncertain time series data streams require significantly more space. Therefore, it is important to construct compressed representation for storing uncertain time series data. Based on granules, sequential sketches are created to store hash-compressed granules. And based on sliding windows, a sketch update strategy is given to store most resent granules. As the sequential sketches may be saturated with the increasing of data streams, this paper presents an optimization strategy to delete the absolute sparse patterns. Based on the sequential sketches, a sequential pattern mining algorithm is proposed for mining uncertain data streams. The experimental results illustrate the effectiveness of the pattern mining algorithm.

DRSP : Dimension Reduction for Similarity Matching and Pruning of Time Series Data Streams

Similarity matching and join of time series data streams has gained a lot of relevance in today's world that has large streaming data. This process finds wide scale application in the areas of location tracking, sensor networks, object positioning and monitoring to name a few. However, as the size of the data stream increases, the cost involved to retain all the data in order to aid the process of similarity matching also increases. We develop a novel framework to addresses the following objectives. Firstly, Dimension reduction is performed in the preprocessing stage, where large stream data is segmented and reduced into a compact representation such that it retains all the crucial information by a technique called Multi-level Segment Means (MSM). This reduces the space complexity associated with the storage of large time-series data streams. Secondly, it incorporates effective Similarity Matching technique to analyze if the new data objects are symmetric to the existing data stream. And finally, the Pruning Technique that filters out the pseudo data object pairs and join only the relevant pairs. The computational cost for MSM is O(l*ni) and the cost for pruning is O(DRF*wsize*d), where DRF is the Dimension Reduction Factor. We have performed exhaustive experimental trials to show that the proposed framework is both efficient and competent in comparison with earlier works.

Indexable online time series segmentation with error bound guarantee

The volume of time series stream data grows rapidly in various applications. To reduce the storage, transmission and processing costs of time series data, segmentation and approximation is a common approach. In this paper, we propose a novel online segmentation algorithm that approximates time series by a set of different types of candidate functions (polynomials of different orders, exponential functions, etc.) and adaptively chooses the most compact one as the pattern of the time series changes. We call this algorithm the Adaptive Approximation (AA) algorithm. The AA algorithm incrementally narrows the feasible coefficient spaces (FCS) of candidate functions in coefficient coordinate systems to make each segment as long as possible given an error bound on each data point. We propose an algorithm called the FCS algorithm for the incremental computation of the feasible coefficient spaces. We further propose a mapping based index for similarity searches on the approximated time series. Experimental results show that our AA algorithm generates more compact approximations of the time series with lower average errors than the state-of-the-art algorithm, and our indexing method processes similarity searches on the approximated time series efficiently.

A Framework to Debug Diagnostic Matrices


 
 
 Diagnostics is an important concept in system health and monitoring of space operations. Many of the existing diagnostic algorithms utilize system knowledge in the form of diagnostic matrix (D-matrix, also popularly known as diagnostic dictionary, fault signature matrix or reachability matrix). The D-matrix maps tests on observed conditions to failures. This matrix is mostly gleaned from physical models during system development. But, sometimes, this may not be enough to obtain high diagnostic performance during operation due to system modifications and lag and noise in sensor measurements. In such a case, it is important to modify this D-matrix based on knowledge obtained from sources such as time-series data stream (simulated or maintenance data) within a framework that includes the diagnostic/inference algorithm. A systematic and sequential update procedure, diagnostic modeling evaluator (DME) is proposed to modify D-matrix and wrapper/test logic considering the least expensive update first. The user sets the diagnostic performance criteria. This iterative procedure includes conditions ranging from modifying 0’s and 1’s in the matrix, adding/removing the rows (failure sources)/columns (tests), or modifying test/wrapper logic used to determine test results. We will experiment this framework on ADAPT datasets from DX challenge 2009.
 
 

High Frequency Statistical Arbitrage Via the Optimal Thermal Causal Path

We consider the problem of identifying similarities and causality relationships in a given set of financial time series data streams. We develop further the “Optimal Thermal Causal Path” method, which is a non-parametric method proposed by Sornette et al. The method considers the mismatch between a given pair of time series in order to identify the expected minimum energy path lead-lag structure between the pair. Traders may find this a useful tool for directional trading, to spot arbitrage opportunities. We add a curvature energy term to the method and we propose an approximation technique to reduce the computational time. We apply the method and approximation technique on various market sectors of NYSE data and extract the highly correlated pairs of time series. We show how traders could exploit arbitrage opportunities by using the method.

Inferential Precision in Single-Case Time-Series Data Streams: How Well Does the EM Procedure Perform When Missing Observations Occur in Autocorrelated Data?

The case-based time-series design is a viable methodology for treatment outcome research. However, the literature has not fully addressed the problem of missing observations with such autocorrelated data streams. Mainly, to what extent do missing observations compromise inference when observations are not independent? Do the available missing data replacement procedures preserve inferential integrity? Does the extent of autocorrelation matter? We use Monte Carlo simulation modeling of a single-subject intervention study to address these questions. We find power sensitivity to be within acceptable limits across four proportions of missing observations (10%, 20%, 30%, and 40%) when missing data are replaced using the Expectation-Maximization Algorithm, more commonly known as the EM Procedure (Dempster, Laird, & Rubin, 1977). This applies to data streams with lag-1 autocorrelation estimates under 0.80. As autocorrelation estimates approach 0.80, the replacement procedure yields an unacceptable power profile. The implications of these findings and directions for future research are discussed.