Suresh C. Kothari

Generalized linear threshold scheme

A generalized linear threshold scheme is introduced. The new scheme generalizes the existing linear threshold schemes. The basic principles involved in the construction of linear threshold schemes are laid out and the relationships between the existing schemes are completely established. The generalized linear scheme is used to provide a hierarchical threshold scheme which allows multiple thresholds necessary in a hierarchical environment.

Adaptation of the relaxation method for learning in bidirectional associative memory

An iterative learning algorithm called PRLAB is described for the discrete bidirectional associative memory (BAM). Guaranteed recall of all training pairs is ensured by PRLAB. The proposed algorithm is significant in many ways. Unlike many existing iterative learning algorithms, PRLAB is not based on the gradient descent technique. It is a novel adaptation from the well-known relaxation method for solving a system of linear inequalities. The algorithm is very fast. Learning 200 random patterns in a 200-200 BAM takes only 20 epochs on the average. PRLAB is highly insensitive to learning parameters and the initial configuration of a BAM. It also offers high scalability for large applications by providing the same high performance when the number of training patterns are increased in proportion to the size of the BAM. An extensive performance analysis of the new learning algorithm is included.

Space and time efficient parallel algorithms and software for EST clustering

Expressed sequence tags, abbreviated as ESTs, are DNA molecules experimentally derived from expressed portions of genes. Clustering of ESTs is essential for gene recognition and for understanding important genetic variations such as those resulting in diseases. We present the algorithmic foundations and implementation of PaCE, a parallel software system we developed for large-scale EST clustering. The novel features of our approach include 1) design of space-efficient algorithms to limit the space required to linear in the size of the input data set, 2) a combination of algorithmic techniques to reduce the total work without sacrificing the quality of EST clustering, and 3) use of parallel processing to reduce runtime and facilitate clustering of large data sets. Using a combination of these techniques, we report the clustering of 327,632 rat ESTs in 47 minutes, and 420,694 Triticum aestivum ESTs in 3 hours and 15 minutes, using a 60-processor IBM xSeries cluster. These problems are well beyond the capabilities of state-of-the-art sequential software. We also present thorough experimental evaluation of our software including quality assessment using benchmark Arabidopsis EST data.

Atlas: a new way to explore software, build analysis tools

Atlas is a new software analysis platform from EnSoft Corp. Atlas decouples the domain-specific analysis goal from its underlying mechanism by splitting analysis into two distinct phases. In the first phase, polynomial-time static analyzers index the software AST, building a rich graph database. In the second phase, users can explore the graph directly or run custom analysis scripts written using a convenient API. These features make Atlas ideal for both interaction and automation. In this paper, we describe the motivation, design, and use of Atlas. We present validation case studies, including the verification of safe synchronization of the Linux kernel, and the detection of malware in Android applications. Our ICSE 2014 demo explores the comprehension and malware detection use cases. Video: http://youtu.be/cZOWlJ-IO0k

The Kappa network with fault-tolerant destination tag algorithm

The design of the Gamma network is analyzed in terms of block structure. The analysis reveals the asymmetry of its duplicate links, and an alternate design in the form of the Kappa network is proposed. Its novel feature is the symmetry of duplicate links at the block level. This symmetry results in a simple control algorithm and enhanced fault tolerance. The relationship between the Kappa network and other existing fault-tolerant networks is briefly discussed. >

A new learning approach to enhance the storage capacity of the Hopfield model

A new learning technique is introduced to solve the problem of the small and restrictive storage capacity of the Hopfield model. The technique exploits the maximum storage capacity. It fails only if appropriate weights do not exist to store the given set of patterns. The technique is not based on the concept of function minimization. Thus, there is no danger of getting stuck in local minima. The technique is free from the step size and moving target problems. Learning speed is very fast and depends on difficulties presented by the training patterns and not so much on the parameters of the algorithm. The technique is scalable. Its performance does not degrade as the problem size increases. An extensive analysis of the learning technique is provided through simulation results.<<ETX>>

Neural Networks for Pattern Recognition

Publisher Summary This chapter provides an account of different neural network architectures for pattern recognition. A neural network consists of several simple processing elements called neurons. Each neuron is connected to some other neurons and possibly to the input nodes. Neural networks provide a simple computing paradigm to perform complex recognition tasks in real time. The chapter categorizes neural networks into three types: single-layer networks, multilayer feedforward networks, and feedback networks. It discusses the gradient descent and the relaxation method as the two underlying mathematical themes for deriving learning algorithms. A lot of research activity is centered on learning algorithms because of their fundamental importance in neural networks. The chapter discusses two important directions of research to improve learning algorithms: the dynamic node generation, which is used by the cascade correlation algorithm; and designing learning algorithms where the choice of parameters is not an issue. It closes with the discussion of performance and implementation issues.

Multistage interconnection networks for multiprocessor systems

Publisher Summary The chapter provides a survey of multistage interconnection networks (MINs) emphasizing the underlying topological design principles. An MIN consists of a sequence of switching stages, each of which consists of several switches. The switching stages are connected with interstage links between successive stages. The chapter discusses the design parameters of MINs. MIN design consists of three different layers: the network topology, functional characteristics of a switch, and the control strategy. These three layers of MIN design are not independent, they strongly influence each other. The chapter examines four important classes of MINs: non-blocking, re-arrangeable, blocking, and multipath. It deals with the performance analysis and very-large-scale implementation (VLSI) of MINs. Combinatorial power (CP), path blockage (PB), and bandwidth (BW) are discussed as three different measures of performance of a MIN. The omega network is presented as a typical example of a network with digit control. The destination tag algorithm, data-alignment requirement, and routing conflicts for the omega network are discussed.

Testing arbitrary subhypergraphs for the lossless join property

Abstract The properties required for an arbitrary subhypergraph to define an embedded join dependency are investigated. The nature of the functional dependencies required to insure the validity of the embedded join dependency is examined. An algorithm for testing an arbitrary subhypergraph for the losslessness property is given for functional dependencies embodied in the edges of the subhypergraph.

Statically-Informed Dynamic Analysis Tools to Detect Algorithmic Complexity Vulnerabilities

Algorithmic Complexity (AC) vulnerabilities can be exploited to cause a denial of service attack. Specifically, an adversary can design an input to trigger excessive (space/time) resource consumption. It is not possible to build a fully automated tool to detect AC vulnerabilities. Since it is an open-ended problem, a human-in-loop exploration is required to find the program loops that could have AC vulnerabilities. Ascertaining whether an arbitrary loop has an AC vulnerability is itself difficult, which is equivalent to the halting problem. This paper is about a pragmatic engineering approach to detect AC vulnerabilities. It presents a statically-informed dynamic (SID) analysis and two tools that provide critical capabilities for detecting AC vulnerabilities. The first is a static analysis tool for exploring the software to find loops as the potential candidates for AC vulnerabilities. The second is a dynamic analysis tool that can try many different inputs to evaluate the selected loops for excessive resource consumption. The two tools are built and integrated together using the interactive software analysis, transformation, and visualization capabilities provided by the Atlas platform. The paper describes two use cases for the tools, one to detect AC vulnerabilities in Java bytecode and another for students in an undergraduate algorithm class to perform experiments to learn different aspects of algorithmic complexity Tool and Demo Video: https://ensoftcorp.github.io/SID.