We propose a resource-efficient Bayesian Optimization (BO) formulation that can provide the same convergence guarantees as traditional BO, while ensuring that the opti-mization makes efficient use of the available cloud or high-performance computing (HPC) resources. The paper is motivated by the fact that for many optimization problems that lend themselves well to BO, like hyper-parameter optimization for training large machine learning models, the single function evaluation cost depends on the model parameters as well as system parameters. The proposed Resource Efficient Bayesian Optimization (REBO) algorithm is a novel formulation that exploits this dependence and provides significant cost benefits for users who want to deploy BO on cloud and HPC resources that are characterized by availability of compute resources with varying costs and expected performance benefits. We demonstrate the effectiveness of REBO, in terms of convergence and resource-efficiency, on a variety of machine learning hyper-parameter optimization applications.

Image classifiers work effectively when applied on structured images, yet they often fail when applied on images with very high visual complexity. This paper describes experiments applying state-of-the-art object classifiers toward a unique set of ‘images in the wild’ with high visual complexity collected on the island of Bali. The text describes differences between actual images in the wild and images from Imagenet, and then discusses a novel approach combining informational cues particular to video with an ensemble of imperfect classifiers in order to improve classification results on video sourced images of plants in the wild.

Analysis of morphological data is central to a broad class of scientific problems in materials science, astronomy, bio-medicine, and many others. Understanding relationships between morphologies is a core analytical task in such settings. In this paper, we propose a graph-based framework for measuring similarity between morphologies. Our framework delivers a novel representation of a morphology as an augmented graph that encodes application-specific knowledge through the use of configurable signature functions. It provides also an algorithm to compute the similarity between a pair of morphology graphs. We present experimental results in which the framework is applied to morphology data from high-fidelity numerical simulations that emerge in materials science. The results demonstrate that our proposed measure is superior in capturing the semantic similarity between morphologies, compared to the state-of-the-art methods such as FFT-based measures.

Scientific and engineering processes deliver massive high-dimensional data sets that are generated as non-linear transformations of an initial state and few process parameters. Mapping such data to a low-dimensional manifold facilitates better understanding of the underlying processes, and enables their optimization. In this paper, we first show that off-theshelf non-linear spectral dimensionality reduction methods, e.g., Isomap, fail for such data, primarily due to the presence of strong temporal correlations. Then, we propose a novel method, Entropy-Isomap, to address the issue. The proposed method is successfully applied to large data describing a fabrication process of organic materials. The resulting low-dimensional representation correctly captures process control variables, allows for low-dimensional visualization of the material morphology evolution, and provides key insights to improve the process.

High performance computing (HPC) facilities consist of a large number of interconnected computing units (or nodes) that execute highly complex scientific simulations to support scientific research. Monitoring such facilities, in real-time, is essential to ensure that the system operates at peak efficiency. Such systems are typically monitored using a variety of measurements and log data which capture the state of the various components within the system at regular intervals of time. As modern HPC systems grow in capacity and complexity, the data produced by current resource monitoring tools at a scale that is no longer feasible to be visually monitored by analysts. We propose a method that transforms the multidimensional output of resource monitoring tools to a low dimensional representation that facilitates the understanding of the behavior of a High Performance Computing (HPC) system. The proposed method automatically extracts the low-dimensional signal in the data which can be used to track the system efficiency and identify performance anomalies. The method models the resource usage data as a three dimensional tensor (capturing resource usage of all compute nodes for different resources over time). A dynamic matrix factorization algorithm, called dynamicMF, is proposed to extract a low-dimensional temporal signal for each node, which is subsequently fed into an anomaly detector. Results on resource usage data show anomalies identified which are correlated with anomalous events identified over the syslog messages.

Analyzing database access logs is a key part of performance tuning, intrusion detection, benchmark development, and many other database administration tasks. Unfortunately, it is common for production databases to deal with millions or more queries each day, so these logs must be summarized before they can be used. Designing an appropriate summary encoding requires trading off between conciseness and information content. For example: simple workload sampling may miss rare, but high impact queries. In this paper, we present L OG R, a lossy log compression scheme suitable for use in many automated log analytics tools, as well as for human inspection. We formalize and analyze the space/fidelity trade-off in the context of a broader family of "pattern" and "pattern mixture" log encodings to which L OG R belongs. We show through a series of experiments that L OG R compressed encodings can be created efficiently, come with provable information-theoretic bounds on their accuracy, and outperform state-of-art log summarization strategies.

Database access logs are the starting point for many forms of database administration, from database performance tuning, to security auditing, to benchmark design, and many more. Unfortunately, query logs are also large and unwieldy, and it can be difficult for an analyst to extract broad patterns from the set of queries found therein. Clustering is a natural first step towards understanding the massive query logs. However, many clustering methods rely on the notion of pairwise similarity, which is challenging to compute for SQL queries, especially when the underlying data and database schema is unavailable. We investigate the problem of computing similarity between queries, relying only on the query structure. We conduct a rigorous evaluation of three query similarity heuristics proposed in the literature applied to query clustering on multiple query log datasets, representing different types of query workloads. To improve the accuracy of the three heuristics, we propose a generic feature engineering strategy, using classical query rewrites to standardize query structure. The proposed strategy results in a significant improvement in the performance of all three similarity heuristics.

Manifold learning based methods have been widely used for non-linear dimensionality reduction (NLDR). However, in many practical settings, the need to process streaming data is a challenge for such methods, owing to the high computational complexity involved. Moreover, most methods operate under the assumption that the input data is sampled from a single manifold, embedded in a high dimensional space. We propose a method for streaming NLDR when the observed data is either sampled from multiple manifolds or irregularly sampled from a single manifold. We show that existing NLDR methods, such as Isomap, fail in such situations, primarily because they rely on smoothness and continuity of the underlying manifold, which is violated in the scenarios explored in this paper. However, the proposed algorithm is able to learn effectively in presence of multiple, and potentially intersecting, manifolds, while allowing for the input data to arrive as a massive stream.

Resource usage data, collected using tools such as TACC_Stats, capture the resource utilization by nodes within a high performance computing system. We present methods to analyze the resource usage data to understand the system performance and identify performance anomalies. The core idea is to model the data as a three-way tensor corresponding to the compute nodes, usage metrics, and time. Using the reconstruction error between the original tensor and the tensor reconstructed from a low rank tensor decomposition, as a scalar performance metric, enables us to monitor the performance of the system in an online fashion. This error statistic is then used for anomaly detection that relies on the assumption that the normal/routine behavior of the system can be captured using a low rank approximation of the original tensor. We evaluate the performance of the algorithm using information gathered from system logs and show that the performance anomalies identified by the proposed method correlates with critical errors reported in the system logs. Results are shown for data collected for 2013 from the Lonestar4 system at the Texas Advanced Computing Center (TACC).

Spectral dimensionality reduction is frequently used to identify low-dimensional structure in high-dimensional data. However, learning manifolds, especially from the streaming data, is computationally and memory expensive. In this paper, we argue that a stable manifold can be learned using only a fraction of the stream, and the remaining stream can be mapped to the manifold in a significantly less costly manner. Identifying the transition point at which the manifold is stable is the key step. We present error metrics that allow us to identify the transition point for a given stream by quantitatively assessing the quality of a manifold learned using Isomap. We further propose an efficient mapping algorithm, called S-Isomap, that can be used to map new samples onto the stable manifold. We describe experiments on a variety of data sets that show that the proposed approach is computationally efficient without sacrificing accuracy.

Generative models for graphs are increasingly becoming a popular tool for researchers to generate realistic approximations of graphs. While in the past, focus was on generating graphs which follow general laws, such as the power law for degree distribution, current models have the ability to learn from observed graphs and generate synthetic approximations. The primary emphasis of existing models has been to closely match different properties of a single observed graph. Such models, though stochastic, tend to generate samples which do not have significant variance in terms of the various graph properties. We argue that in many cases real graphs are sampled drawn from a graph population (e.g., networks sampled at various time points, social networks for individual schools, healthcare networks for different geographic regions, etc.). Such populations typically exhibit significant variance. However, existing models are not designed to model this variance, which could lead to issues such as overfitting. We propose a graph generative model that focuses on matching the properties of real graphs and the natural variance expected for the corresponding population. The proposed model adopts a mixture-model strategy to expand the expressiveness of Kronecker product based graph models (KPGM), while building upon the two strengths of KPGM, viz., ability to model several key properties of graphs and to scale to massive graph sizes using its elegant fractal growth based formulation. The proposed model, called x-Kronecker Product Graph Model, or xKPGM, allows scalable learning from observed graphs and generates samples that match the mean and variance of several salient graph properties. We experimentally demonstrate the capability of the proposed model to capture the inherent variability in real world graphs on a variety of publicly available graph data sets.

he role of big data in addressing the needs of the present healthcare system in US and rest of the world has been echoed by government, private, and academic sectors. There has been a growing emphasis to explore the promise of big data analytics in tapping the potential of the massive healthcare data emanating from private and government health insurance providers. While the domain implications of such collaboration are well known, this type of data has been explored to a limited extent in the data mining community. The objective of this paper is two fold: first, we introduce the emerging domain of "big" healthcare claims data to the KDD community, and second, we describe the success and challenges that we encountered in analyzing this data using state of art analytics for massive data. Specifically, we translate the problem of analyzing healthcare data into some of the most well-known analysis problems in the data mining community, social network analysis, text mining, and temporal analysis and higher order feature construction, and describe how advances within each of these areas can be leveraged to understand the domain of healthcare. Each case study illustrates a unique intersection of data mining and healthcare with a common objective of improving the cost-care ratio by mining for opportunities to improve healthcare operations and reducing what seems to fall under fraud, waste, and abuse.

This survey attempts to provide a comprehensive and structured overview of the existing research for the problem of detecting anomalies in discrete/symbolic sequences. The objective is to provide a global understanding of the sequence anomaly detection problem and how existing techniques relate to each other. The key contribution of this survey is the classification of the existing research into three distinct categories, based on the problem formulation that they are trying to solve. These problem formulations are: 1) identifying anomalous sequences with respect to a database of normal sequences; 2) identifying an anomalous subsequence within a long sequence; and 3) identifying a pattern in a sequence whose frequency of occurrence is anomalous. We show how each of these problem formulations is characteristically distinct from each other and discuss their relevance in various application domains. We review techniques from many disparate and disconnected application domains that address each of these formulations. Within each problem formulation, we group techniques into categories based on the nature of the underlying algorithm. For each category, we provide a basic anomaly detection technique, and show how the existing techniques are variants of the basic technique. This approach shows how different techniques within a category are related or different from each other. Our categorization reveals new variants and combinations that have not been investigated before for anomaly detection. We also provide a discussion of relative strengths and weaknesses of different techniques. We show how techniques developed for one problem formulation can be adapted to solve a different formulation, thereby providing several novel adaptations to solve the different problem formulations. We also highlight the applicability of the techniques that handle discrete sequences to other related areas such as online anomaly detection and time series anomaly detection.

Accurate damage assessment due to major natural and anthropogenic disasters is becoming critical due to increasing human and economic losses. This increase in loss of life and severe damages can be attributed to the growing population, and human migration and settlements in disaster prone regions of the world. Rapid damage assessment and dissemination of accurate information is critical for creating an effective emergency response. Remote sensing and geographic information systems (GIS) based techniques and tools are playing an important function in disaster damage assessment and reporting activities. In this review, we will look into the current state of art, in damage assessment using remote sensing and GIS based techniques.

Online time series change detection is a critical component of many monitoring systems, such as space and air-borne remote sensing instruments, cardiac monitors, and network traffic profilers, which continuously analyze observations recorded by sensors. Data collected by such sensors typically has a periodic component. Most existing time series change detection methods are not directly applicable to handle such data, either because they are not designed to handle periodic time series or because they cannot operate in an online mode. We propose an online change detection algorithm which can handle periodic time series. The algorithm uses a Gaussian process based non-parametric time series prediction model and monitors the difference between the predictions and actual observations within a statistical control chart framework to identify changes. A key challenge in using Gaussian process in an online mode is the need to solve a large system of equations involving the associated covariance matrix which grows with every time step. The proposed algorithm exploits the special structure of the covariance matrix and can analyze a time series of length T in O(T2) time while maintaining a O(T) memory footprint, compared to O(T4) time and O(T2) memory requirement of standard matrix manipulation methods. We experimentally demonstrate the superiority of the proposed algorithm over several existing time series change detection algorithms on a set of synthetic and real time series. Finally, we illustrate the effectiveness of the proposed algorithm for identifying land use land cover changes using Normalized Difference Vegetation Index (NDVI) data collected for an agricultural region in Iowa state, USA. Our algorithm is able to detect different types of changes in a NDVI validation data set (with ≈ 80% accuracy) which occur due to crop type changes as well as natural disasters.

In this paper, we present a framework for categorical data analysis which allows such data sets to be explored using a rich set of techniques that are only applicable to continuous data sets. We introduce the concept of separability statistics in the context of exploratory categorical data analysis. We show how these statistics can be used as a way to map categorical data to continuous space given a labeled reference data set. This mapping enables visualization of categorical data using techniques that are applicable to continuous data. We show that in the transformed continuous space, the performance of the standard k-nn based outlier detection technique is comparable to the performance of the k-nn based outlier detection technique using the best of the similarity measures designed for categorical data. The proposed framework can also be used to devise similarity measures best suited for a particular type of data set.

We present a comparative evaluation of a large number of anomaly detection techniques on a variety of publicly available as well as artificially generated data sets. Many of these are existing techniques while some are slight variants and/or adaptations of traditional anomaly detection techniques to sequence data.

Measuring similarity or distance between two entities is a key step for several data mining and knowledge discovery tasks. The notion of similarity for continuous data is relatively well-understood, but for categorical data, the similarity computation is not straightforward. Several data-driven similarity measures have been proposed in the literature to compute the similarity between two categorical data instances but their relative performance has not been evaluated. In this paper we study the performance of a variety of similarity measures in the context of a specific data mining task: outlier detection. Results on a variety of data sets show that while no one measure dominates others for all types of problems, some measures are able to have consistently high performance.

In this paper, we formulate the problem of summarization of a dataset of transactions with categorical attributes as an optimization problem involving two objective functions - compaction gain and information loss. We propose metrics to characterize the output of any summarization algorithm. We investigate two approaches to address this problem. The first approach is an adaptation of clustering and the second approach makes use of frequent item sets from the association analysis domain. We illustrate one application of summarization in the field of network data where we show how our technique can be effectively used to summarize network traffic into a compact but meaningful representation. Specifically, we evaluate our proposed algorithms on the 1998 DARPA Off-line Intrusion Detection Evaluation data and network data generated by SKAION Corp for the ARDA information assurance program.