Karthik Ganesan Pillai

A large-scale solar image dataset with labeled event regions

This paper introduces a new public benchmark dataset of solar image data from the Solar Dynamics Observatory (SDO) mission. This is the first release, which contains over 15,000 images and nearly 24,000 solar events, spanning the first six months of 2012. It combines region-based event labels from six automated detection modules, ten pre-computed image parameters for each cell over a grid-based segmentation of the full resolution images, and a lower resolution version of the images for further analysis and visualization. Together, these components serve as a standardized, ready-to-use, solar image dataset for general image processing research, without requiring the necessary background knowledge to properly prepare it. We present here the fundamental dataset creation details and outline future improvements and opportunities as data collection continues for the coming years.

A filter-and-refine approach to mine spatiotemporal co-occurrences

Spatiotemporal co-occurrence patterns (STCOPs) represent the subsets of event types that occur together in both space and time. However, the discovery of STCOPs in data sets with extended spatial representations that evolve over time is computationally expensive because of the necessity to calculate interest measures to assess the co-occurrence strength, and the number of candidates for STCOPs growing exponentially with the number of spatiotemporal event types. In this paper, we introduce a novel and effective filter-and-refine algorithm to efficiently find prevalent STCOPs in massive spatiotemporal data repositories with polygon shapes that move and evolve over time. We provide theoretical analysis of our approach, and follow this investigation with a practical evaluation of our algorithm effectiveness on three real-life data sets and one artificial data set.

Spatio-temporal Co-occurrence Pattern Mining in Data Sets with Evolving Regions

Spatio-temporal co-occurring patterns represent subsets of event types that occur together in both space and time. In comparison to previous work in this field, we present a general framework to identify spatio-temporal co occurring patterns for continuously evolving spatio-temporal events that have polygon-like representations. We also propose a set of measures to identify spatio-temporal co-occurring patterns and propose an Apriori-based spatio-temporal co-occurrence mining algorithm to find prevalent spatio-temporal co-occurring patterns for extended spatial representations that evolve over time. We evaluate our framework on real-life data to demonstrate the effectiveness of our measures and the algorithm. We present results highlighting the importance of our measures in identifying spatio-temporal co-occurrence patterns.

Spatiotemporal indexing techniques for efficiently mining spatiotemporal co-occurrence patterns

In this paper, we investigate using specifically-designated spatiotemporal indexing techniques for mining cooccurrence patterns from spatiotemporal datasets with evolving polygon-based representations. Previously, suggested techniques for spatiotemporal pattern mining algorithms did not take spatiotemporal indexing techniques into account. We present a new framework for mining spatiotemporal co-occurrence patterns that can use various indexing techniques for efficiently accessing data. Two well-studied spatiotemporal indexing structures, Scalable and Efficient Trajectory Index (SETI) and Chebyshev Polynomial Indexing are currently implemented and available in our framework.

Overlapping swarm intelligence for training artificial neural networks

A novel overlapping swarm intelligence algorithm is introduced to train the weights of an artificial neural network. Training a neural network is a difficult task that requires an effective search methodology to compute the weights along the edges of a network. The backpropagation algorithm, a gradient based method, is frequently used to train multilayer feed-forward networks. Gradient based methods might not always lead to a globally optimal solution of the network. On the other hand, training algorithms based on evolutionary computation have been used to train multilayer feed-forward networks in an attempt to overcome the limitations of gradient based algorithms with mixed results. This paper introduces an overlapping swarm intelligence technique to train multilayer feedforward networks. The results show that OSI method performs either on par with or better than the other methods tested.

DOSI: Training artificial neural networks using overlapping swarm intelligence with local credit assignment

A novel swarm-based algorithm is proposed for the training of artificial neural networks. Training of such networks is a difficult problem that requires an effective search algorithm to find optimal weight values. While gradient-based methods, such as backpropagation, are frequently used to train multilayer feedforward neural networks, such methods may not yield a globally optimal solution. To overcome the limitations of gradient-based methods, evolutionary algorithms have been used to train these networks with some success. This paper proposes an overlapping swarm intelligence algorithm for training neural networks in which a particle swarm is assigned to each neuron to search for that neurons weights. Unlike similar architectures, our approach does not require a shared global network for fitness evaluation. Thus the approach discussed in this paper localizes the credit assignment process by first focusing on updating weights within local swarms and then evaluating the fitness of the particles using a localized network. This has the advantage of enabling our algorithms learning process to be fully distributed.

Spatiotemporal Co-occurrence Rules

Spatiotemporal co-occurrence rules (STCORs) discovery is an important problem in many application domains such as weather monitoring and solar physics, which is our application focus. In this paper, we present a general framework to identify STCORs for continuously evolving spatiotemporal events that have extended spatial representations. We also analyse a set of anti-monotone (monotonically non-increasing) and non anti-monotone measures to identify STCORs. We then validate and evaluate our framework on a real-life data set and report results of the comparison of the number candidates needed to discover actual patterns, memory usage, and the number of STCORs discovered using the anti-monotonic and non anti-monotonic measures.

Iterative refinement of multiple targets tracking of solar events

In this paper, we combine two approaches to multiple-target tracking: the first is a hierarchical approach to iteratively growing track fragments across gaps in detections, and the second is a network flow based optimization method for data association. We introduce a new parallel algorithm for initial track fragment formation as the base of the hierarchical approach. The network flow based optimization method is then utilized for the remaining levels of the hierarchy. This process is applied to solar data retrieved from the Heliophysics Event Knowledgebase (HEK). We compare our results to labeled data from the same, and show improvements over a non-hierarchical sequential approach.

Big Data New Frontiers: Mining, Search and Management of Massive Repositories of Solar Image Data and Solar Events

This work presents one of the many emerging research domains where big data analysis has become an immediate need to process the massive amounts of data being generated each day: solar physics. While building a content-based image retrieval system for NASA’s Solar Dynamics Observatory mission, we have discovered research problems that can be addressed by the use of big data processing techniques and in some cases require the development of novel techniques. With over one terabyte of solar data being generated each day, and ever more missions on the horizon that expect to generate petabytes of data each year, solar physics presents many exciting opportunities. This paper presents the current status of our work with solar image data and events, our shift towards using big data methodologies, and future directions for big data processing in solar physics.

Multi-sensor Remote Sensing Image Change Detection: An Evaluation of Similarity Measures

Change detection from remote sensing imagery is of great interest in disaster management, surveillance, and other applications. Most of the existing approaches are pixel based and rely on direct comparison of radiometric values to detect changes. Such techniques are susceptible to atmospheric conditions, noise, and registration errors. In this paper, we evaluate change detection approaches using several similarity measures that does not rely entirely on radiometric values of the images. Our hypothesis is based on the assumption, that even though images are obtained from different sensors and at different times, the underlying basis in the scene is still the same, since they are different representations of the same reality. In other words, different sensors capture overlapping information in different forms. Thus, we expect that similarity measures provides contrasting information for change vs. no change patches. We evaluated several measures and experimental results show the effectiveness of each measure in identifying the changed regions in the images.