arXiv subjects

Sanjay Chawla

Publications and source records attributed to Sanjay Chawla.

68 records · Page 4Linked to original sources

arXiv2016

Scalable Constrained Clustering: A Generalized Spectral Method

We present a simple spectral approach to the well-studied constrained clustering problem. It captures constrained clustering as a generalized eigenvalue problem with graph Laplacians. The algorithm works in nearly-linear time and provides concrete guarantees for the quality of the clusters, at least for the case of 2-way partitioning. In practice this translates to a very fast implementation that consistently outperforms existing spectral approaches both in speed and quality.

cs.SI↗

arXiv2015

Discovering the Network Backbone from Traffic Activity Data

We introduce a new computational problem, the BackboneDiscovery problem, which encapsulates both functional and structural aspects of network analysis. While the topology of a typical road network has been available for a long time (e.g., through maps), it is only recently that fine-granularity functional (activity and usage) information about the network (like source-destination traffic information) is being collected and is readily available. The combination of functional and structural information provides an efficient way to explore and understand usage patterns of networks and aid in design and decision making. We propose efficient algorithms for the BackboneDiscovery problem including a novel use of edge centrality. We observe that for many real world networks, our algorithm produces a backbone with a small subset of the edges that support a large percentage of the network activity.

cs.SI↗

arXiv2015

Parameter Database : Data-centric Synchronization for Scalable Machine Learning

We propose a new data-centric synchronization framework for carrying out of machine learning (ML) tasks in a distributed environment. Our framework exploits the iterative nature of ML algorithms and relaxes the application agnostic bulk synchronization parallel (BSP) paradigm that has previously been used for distributed machine learning. Data-centric synchronization complements function-centric synchronization based on using stale updates to increase the throughput of distributed ML computations. Experiments to validate our framework suggest that we can attain substantial improvement over BSP while guaranteeing sequential correctness of ML tasks.

cs.DB↗

arXiv2015

On inferring structural connectivity from brain functional-MRI data

The anatomical structure of the brain can be observed via non-invasive techniques such as diffusion imaging. However, these are imperfect because they miss connections that are actually known to exist, especially long range inter-hemispheric ones. In this paper we formulate the inverse problem of inferring the structural connectivity of brain networks from experimentally observed functional connectivity via functional Magnetic Resonance Imaging (fMRI), by formulating it as a convex optimization problem. We show that structural connectivity can be modeled as an optimal sparse representation derived from the much denser functional connectivity in the human brain. Using only the functional connectivity data as input, we present (a) an optimization problem that models constraints based on known physiological observations, and (b) an ADMM algorithm for solving it. The algorithm not only recovers the known structural connectivity of the brain, but is also able to robustly predict the long range inter-hemispheric connections missed by DSI or DTI, including a very good match with experimentally observed quantitative distributions of the weights/strength of anatomical connections. We demonstrate results on both synthetic model data and a fine-scale 998 node cortical dataset, and discuss applications to other complex network domains where retrieving effective structure from functional signatures are important.

q-bio.NC↗

arXiv2014

Classification of Passes in Football Matches using Spatiotemporal Data

A knowledgeable observer of a game of football (soccer) can make a subjective evaluation of the quality of passes made between players during the game. We investigate the problem of producing an automated system to make the same evaluation of passes. We present a model that constructs numerical predictor variables from spatiotemporal match data using feature functions based on methods from computational geometry, and then learns a classification function from labelled examples of the predictor variables. Furthermore, the learned classifiers are analysed to determine if there is a relationship between the complexity of the algorithm that computed the predictor variable and the importance of the variable to the classifier. Experimental results show that we are able to produce a classifier with 85.8% accuracy on classifying passes as Good, OK or Bad, and that the predictor variables computed using complex methods from computational geometry are of moderate importance to the learned classifiers. Finally, we show that the inter-rater agreement on pass classification between the machine classifier and a human observer is of similar magnitude to the agreement between two observers.

cs.LG↗

arXiv2014

Integer Programming Relaxations for Integrated Clustering and Outlier Detection

In this paper we present methods for exemplar based clustering with outlier selection based on the facility location formulation. Given a distance function and the number of outliers to be found, the methods automatically determine the number of clusters and outliers. We formulate the problem as an integer program to which we present relaxations that allow for solutions that scale to large data sets. The advantages of combining clustering and outlier selection include: (i) the resulting clusters tend to be compact and semantically coherent (ii) the clusters are more robust against data perturbations and (iii) the outliers are contextualised by the clusters and more interpretable, i.e. it is easier to distinguish between outliers which are the result of data errors from those that may be indicative of a new pattern emergent in the data. We present and contrast three relaxations to the integer program formulation: (i) a linear programming formulation (LP) (ii) an extension of affinity propagation to outlier detection (APOC) and (iii) a Lagrangian duality based formulation (LD). Evaluation on synthetic as well as real data shows the quality and scalability of these different methods.

cs.LG↗

arXiv2014

Sleep Analytics and Online Selective Anomaly Detection

We introduce a new problem, the Online Selective Anomaly Detection (OSAD), to model a specific scenario emerging from research in sleep science. Scientists have segmented sleep into several stages and stage two is characterized by two patterns (or anomalies) in the EEG time series recorded on sleep subjects. These two patterns are sleep spindle (SS) and K-complex. The OSAD problem was introduced to design a residual system, where all anomalies (known and unknown) are detected but the system only triggers an alarm when non-SS anomalies appear. The solution of the OSAD problem required us to combine techniques from both machine learning and control theory. Experiments on data from real subjects attest to the effectiveness of our approach.

cs.LG↗

arXiv2014

Network Traffic Decomposition for Anomaly Detection

In this paper we focus on the detection of network anomalies like Denial of Service (DoS) attacks and port scans in a unified manner. While there has been an extensive amount of research in network anomaly detection, current state of the art methods are only able to detect one class of anomalies at the cost of others. The key tool we will use is based on the spectral decomposition of a trajectory/hankel matrix which is able to detect deviations from both between and within correlation present in the observed network traffic data. Detailed experiments on synthetic and real network traces shows a significant improvement in detection capability over competing approaches. In the process we also address the issue of robustness of anomaly detection systems in a principled fashion.

cs.LG↗

arXiv2013

Feature Graph Architectures

In this article we propose feature graph architectures (FGA), which are deep learning systems employing a structured initialisation and training method based on a feature graph which facilitates improved generalisation performance compared with a standard shallow architecture. The goal is to explore alternative perspectives on the problem of deep network training. We evaluate FGA performance for deep SVMs on some experimental datasets, and show how generalisation and stability results may be derived for these models. We describe the effect of permutations on the model accuracy, and give a criterion for the optimal permutation in terms of feature correlations. The experimental results show that the algorithm produces robust and significant test set improvements over a standard shallow SVM training method for a range of datasets. These gains are achieved with a moderate increase in time complexity.

cs.LG↗

arXiv2012

Large Scale Spectral Clustering Using Approximate Commute Time Embedding

Spectral clustering is a novel clustering method which can detect complex shapes of data clusters. However, it requires the eigen decomposition of the graph Laplacian matrix, which is proportion to $O(n^3)$ and thus is not suitable for large scale systems. Recently, many methods have been proposed to accelerate the computational time of spectral clustering. These approximate methods usually involve sampling techniques by which a lot information of the original data may be lost. In this work, we propose a fast and accurate spectral clustering approach using an approximate commute time embedding, which is similar to the spectral embedding. The method does not require using any sampling technique and computing any eigenvector at all. Instead it uses random projection and a linear time solver to find the approximate embedding. The experiments in several synthetic and real datasets show that the proposed approach has better clustering quality and is faster than the state-of-the-art approximate spectral clustering methods.

cs.LG↗

arXiv2012

SparseDTW: A Novel Approach to Speed up Dynamic Time Warping

We present a new space-efficient approach, (SparseDTW), to compute the Dynamic Time Warping (DTW) distance between two time series that always yields the optimal result. This is in contrast to other known approaches which typically sacrifice optimality to attain space efficiency. The main idea behind our approach is to dynamically exploit the existence of similarity and/or correlation between the time series. The more the similarity between the time series the less space required to compute the DTW between them. To the best of our knowledge, all other techniques to speedup DTW, impose apriori constraints and do not exploit similarity characteristics that may be present in the data. We conduct experiments and demonstrate that SparseDTW outperforms previous approaches.

cs.DB↗

arXiv2011

An index for regular expression queries: Design and implementation

The like regular expression predicate has been part of the SQL standard since at least 1989. However, despite its popularity and wide usage, database vendors provide only limited indexing support for regular expression queries which almost always require a full table scan. In this paper we propose a rigorous and robust approach for providing indexing support for regular expression queries. Our approach consists of formulating the indexing problem as a combinatorial optimization problem. We begin with a database, abstracted as a collection of strings. From this data set we generate a query workload. The input to the optimization problem is the database and the workload. The output is a set of multigrams (substrings) which can be used as keys to records which satisfy the query workload. The multigrams can then be integrated with the data structure (like B+ trees) to provide indexing support for the queries. We provide a deterministic and a randomized approximation algorithm (with provable guarantees) to solve the optimization problem. Extensive experiments on synthetic data sets demonstrate that our approach is accurate and efficient. We also present a case study on PROSITE patterns - which are complex regular expression signatures for classes of proteins. Again, we are able to demonstrate the utility of our indexing approach in terms of accuracy and efficiency. Thus, perhaps for the first time, there is a robust and practical indexing mechanism for an important class of database queries.

cs.DB↗

arXiv2011

Online Anomaly Detection Systems Using Incremental Commute Time

Commute Time Distance (CTD) is a random walk based metric on graphs. CTD has found widespread applications in many domains including personalized search, collaborative filtering and making search engines robust against manipulation. Our interest is inspired by the use of CTD as a metric for anomaly detection. It has been shown that CTD can be used to simultaneously identify both global and local anomalies. Here we propose an accurate and efficient approximation for computing the CTD in an incremental fashion in order to facilitate real-time applications. An online anomaly detection algorithm is designed where the CTD of each new arriving data point to any point in the current graph can be estimated in constant time ensuring a real-time response. Moreover, the proposed approach can also be applied in many other applications that utilize commute time distance.

cs.AI↗

arXiv2010

Angle Tree: Nearest Neighbor Search in High Dimensions with Low Intrinsic Dimensionality

We propose an extension of tree-based space-partitioning indexing structures for data with low intrinsic dimensionality embedded in a high dimensional space. We call this extension an Angle Tree. Our extension can be applied to both classical kd-trees as well as the more recent rp-trees. The key idea of our approach is to store the angle (the "dihedral angle") between the data region (which is a low dimensional manifold) and the random hyperplane that splits the region (the "splitter"). We show that the dihedral angle can be used to obtain a tight lower bound on the distance between the query point and any point on the opposite side of the splitter. This in turn can be used to efficiently prune the search space. We introduce a novel randomized strategy to efficiently calculate the dihedral angle with a high degree of accuracy. Experiments and analysis on real and synthetic data sets shows that the Angle Tree is the most efficient known indexing structure for nearest neighbor queries in terms of preprocessing and space usage while achieving high accuracy and fast search time.

cs.DS↗