Kannan, Sampath
Overview
Works:  22 works in 28 publications in 1 language and 39 library holdings 

Roles:  Author, Editor 
Publication Timeline
.
Most widely held works by
Sampath Kannan
Special issue on STOC 2001 : [... at the 33rd ACM Symposium on Theory of Computing (STOC), held in Crete, Greece, July 68,
2001] by
2001, Chersonēsos> Symposium on Theory of Computing. <33(
Book
)
3 editions published in 2004 in English and held by 5 WorldCat member libraries worldwide
3 editions published in 2004 in English and held by 5 WorldCat member libraries worldwide
Anchors in tournaments by S Kannan(
Book
)
2 editions published in 1992 in English and held by 5 WorldCat member libraries worldwide
2 editions published in 1992 in English and held by 5 WorldCat member libraries worldwide
On the query complexity of learning and a technique for lower bounds on monotone formulae by S. K Kannan(
Book
)
1 edition published in 1991 in English and held by 3 WorldCat member libraries worldwide
Abstract: "We consider the problem of learning parametrized concept classes with membership and equivalence queries. If C[subscript n] is the concept class being learned, we show that if equivalence queries can be made from a larger but still 'reasonable' hypothesis class, then there exist [formula] queries that exactly learn the target concept c [element of] C[subscript n]. We also show that our results are best possible in terms of how big the hypothesis class needs to be and thereby give a way of deriving a lower bound of [omega](n?) on the size of monotone formulae for majority and other boolean functions. This matches the best known lower bounds for majority."
1 edition published in 1991 in English and held by 3 WorldCat member libraries worldwide
Abstract: "We consider the problem of learning parametrized concept classes with membership and equivalence queries. If C[subscript n] is the concept class being learned, we show that if equivalence queries can be made from a larger but still 'reasonable' hypothesis class, then there exist [formula] queries that exactly learn the target concept c [element of] C[subscript n]. We also show that our results are best possible in terms of how big the hypothesis class needs to be and thereby give a way of deriving a lower bound of [omega](n?) on the size of monotone formulae for majority and other boolean functions. This matches the best known lower bounds for majority."
Program checkers for algebraic problems by Sampath Kumar Kannan(
Book
)
2 editions published in 1989 in English and held by 3 WorldCat member libraries worldwide
2 editions published in 1989 in English and held by 3 WorldCat member libraries worldwide
A probabilistic algorithm for merging two sorted lists by S Kannan(
Book
)
1 edition published in 1989 in English and held by 3 WorldCat member libraries worldwide
1 edition published in 1989 in English and held by 3 WorldCat member libraries worldwide
Program checkers for probability generation by Sampath Kannan(
Book
)
1 edition published in 1991 in English and held by 2 WorldCat member libraries worldwide
1 edition published in 1991 in English and held by 2 WorldCat member libraries worldwide
Program correctness checking ... and the design of programs that check their work by
Manuel Blum(
Book
)
2 editions published in 1988 in English and held by 2 WorldCat member libraries worldwide
2 editions published in 1988 in English and held by 2 WorldCat member libraries worldwide
Designing programs that check their work by Manuel Blum(
Book
)
1 edition published in 1995 in English and held by 1 WorldCat member library worldwide
1 edition published in 1995 in English and held by 1 WorldCat member library worldwide
Content selection in multidocument summarization by Kai Hong(
Book
)
1 edition published in 2015 in English and held by 1 WorldCat member library worldwide
Automatic summarization has advanced greatly in the past few decades. However, there remains a huge gap between the content quality of human and machine summaries. There is also a large disparity between the performance of current systems and that of the best possible automatic systems. In this thesis, we explore how the content quality of machine summaries can be improved. First, we introduce a supervised model to predict the importance of words in the input sets, based on a rich set of features. Our model is superior to prior methods in identifying words used in human summaries (i.e., summary keywords). We show that a modular extractive summarizer using the estimates of word importance can generate summaries comparable to the stateoftheart systems. Among the features we propose, we highlight global knowledge, which estimate word importance based on information independent of the input. In particular, we explore two kinds of global knowledge: (1) important categories mined from dictionaries, and (2) intrinsic importance of words. We show that global knowledge is very useful in identifying summary keywords that have low frequency in the input. Second, we present a new framework of system combination for multidocument summarization. This is motivated by our observation that different systems generate very different summaries. For each input set, we generate candidate summaries by combining whole sentences produced by different systems. We show that the oracle summary among these candidates is much better than the output from the systems that we have combined. We then introduce a support vector regression model to select among these candidates. The features we employ in this model capture the informativeness of a summary based on the input documents, the outputs of different systems, and global knowledge. Our model achieves considerable improvement over the systems that we have combined while generating summaries up to a certain length. Furthermore, we study what factors could affect the success of system combination. Experiments show that it is important for the systems combined to have a similar performance
1 edition published in 2015 in English and held by 1 WorldCat member library worldwide
Automatic summarization has advanced greatly in the past few decades. However, there remains a huge gap between the content quality of human and machine summaries. There is also a large disparity between the performance of current systems and that of the best possible automatic systems. In this thesis, we explore how the content quality of machine summaries can be improved. First, we introduce a supervised model to predict the importance of words in the input sets, based on a rich set of features. Our model is superior to prior methods in identifying words used in human summaries (i.e., summary keywords). We show that a modular extractive summarizer using the estimates of word importance can generate summaries comparable to the stateoftheart systems. Among the features we propose, we highlight global knowledge, which estimate word importance based on information independent of the input. In particular, we explore two kinds of global knowledge: (1) important categories mined from dictionaries, and (2) intrinsic importance of words. We show that global knowledge is very useful in identifying summary keywords that have low frequency in the input. Second, we present a new framework of system combination for multidocument summarization. This is motivated by our observation that different systems generate very different summaries. For each input set, we generate candidate summaries by combining whole sentences produced by different systems. We show that the oracle summary among these candidates is much better than the output from the systems that we have combined. We then introduce a support vector regression model to select among these candidates. The features we employ in this model capture the informativeness of a summary based on the input documents, the outputs of different systems, and global knowledge. Our model achieves considerable improvement over the systems that we have combined while generating summaries up to a certain length. Furthermore, we study what factors could affect the success of system combination. Experiments show that it is important for the systems combined to have a similar performance
Computing the local consensus of trees by Sampath Kannan(
)
1 edition published in 1995 in English and held by 1 WorldCat member library worldwide
1 edition published in 1995 in English and held by 1 WorldCat member library worldwide
Determining the evolutionary tree by Sampath Kannan(
)
1 edition published in 1990 in English and held by 1 WorldCat member library worldwide
1 edition published in 1990 in English and held by 1 WorldCat member library worldwide
A fast algorithm for the computation and enumeration of perfect phylogenies when the number of character states is fixed by Sampath Kannan(
)
1 edition published in 1995 in English and held by 1 WorldCat member library worldwide
1 edition published in 1995 in English and held by 1 WorldCat member library worldwide
Stochastic tree processes by Kevin Tian(
Book
)
1 edition published in 2016 in English and held by 1 WorldCat member library worldwide
The second setting concerns computing in the presence of faults. In pushing the limits of computing hardware, there is tradeoff between the reliability of components and their cost. We first examine a method of identifying faulty gates in a readonce formula when our access is limited to providing an input and reading its output. We show that determining whether a fault exists can always be done, and that locating these faults can be done efficiently as long as the readonce formula satisfies a certain balance condition. Finally for a fixed topology, we provide a dynamic program which allows us to optimize how to allocate resources to individual gates so as to optimize the reliability of the whole system under a known input product distribution
1 edition published in 2016 in English and held by 1 WorldCat member library worldwide
The second setting concerns computing in the presence of faults. In pushing the limits of computing hardware, there is tradeoff between the reliability of components and their cost. We first examine a method of identifying faulty gates in a readonce formula when our access is limited to providing an input and reading its output. We show that determining whether a fault exists can always be done, and that locating these faults can be done efficiently as long as the readonce formula satisfies a certain balance condition. Finally for a fixed topology, we provide a dynamic program which allows us to optimize how to allocate resources to individual gates so as to optimize the reliability of the whole system under a known input product distribution
Register allocation in structured programs by Sampath Kannan(
)
1 edition published in 1995 in English and held by 1 WorldCat member library worldwide
1 edition published in 1995 in English and held by 1 WorldCat member library worldwide
Checkers, SelfTesters, and SelfCorrectors for Reactive Systems(
Book
)
2 editions published in 2001 in English and held by 1 WorldCat member library worldwide
This report discusses the development of formal methods for monitoring safetycritical realtime and reactive systems. This project centers on building on expertise in the area of processalgebrabased specification and analysis of realtime systems as well as the paradigm of program checking which allows one to make rigorous statements about the correctness of program behavior rather than of the program itself. To integrate these ideas a prototype system (JavaMAC) for monitoring and checking Java programs has been implemented. MAC takes a monitoring script provided by the user, the program, and a requirement specification and produces a) an instrumentation of the program to send variable update information to the monitoring and checking unit b) a script for transforming low level program variable values to abstract events and c) a script for checking whether a sequence of events is consistent with the desired property. These scripts written in new languages are defined (PEDL and MEDL respectively) and are then used to produce other components that extract lowlevel information from the program, convert it to events and check that the sequence of events represents correct behavior. The prototype has been successfully tested on two applications  micro air vehicles attaining a desired formation, and convergence of a network routing protocol. Performance measurements have been done on JavaMAC in an attempt to breakdown the overhead introduced by JavaMAC into its various components. Subsequently several optimizations in JavaMAC have been introduced to improve the performance. Other research funded by this grant includes papers on probabilistic bisimulation and on lowoverhead checking of the correctness of the output produced by programs for sorting and other mathematically welldefined' tasks
2 editions published in 2001 in English and held by 1 WorldCat member library worldwide
This report discusses the development of formal methods for monitoring safetycritical realtime and reactive systems. This project centers on building on expertise in the area of processalgebrabased specification and analysis of realtime systems as well as the paradigm of program checking which allows one to make rigorous statements about the correctness of program behavior rather than of the program itself. To integrate these ideas a prototype system (JavaMAC) for monitoring and checking Java programs has been implemented. MAC takes a monitoring script provided by the user, the program, and a requirement specification and produces a) an instrumentation of the program to send variable update information to the monitoring and checking unit b) a script for transforming low level program variable values to abstract events and c) a script for checking whether a sequence of events is consistent with the desired property. These scripts written in new languages are defined (PEDL and MEDL respectively) and are then used to produce other components that extract lowlevel information from the program, convert it to events and check that the sequence of events represents correct behavior. The prototype has been successfully tested on two applications  micro air vehicles attaining a desired formation, and convergence of a network routing protocol. Performance measurements have been done on JavaMAC in an attempt to breakdown the overhead introduced by JavaMAC into its various components. Subsequently several optimizations in JavaMAC have been introduced to improve the performance. Other research funded by this grant includes papers on probabilistic bisimulation and on lowoverhead checking of the correctness of the output produced by programs for sorting and other mathematically welldefined' tasks
Programming using automata and transducers by Loris D'Antoni(
Book
)
1 edition published in 2015 in English and held by 1 WorldCat member library worldwide
First, we introduce succinct models of transducers that can operate over large alphabets and design BEX, a language for analysing string coders. We use BEX to prove the correctness of UTF and B ASE64 encoders and decoders. Next, we develop a theory of tree transducers over infinite alphabets and design FAST, a language for analysing treemanipulating programs. We use FAST to detect vulnerabilities in HTML sanitizers, check whether augmented reality taggers conflict, and optimize and analyze functional programs that operate over lists and trees. Finally, we focus on laying the foundations of stream processing of hierarchical data such as XML files and program traces. We introduce two new efficient and executable models that can process the input in a lefttoright linear pass: symbolic visibly pushdown automata and streaming tree transducers. Symbolic visibly pushdown automata are closed under Boolean operations and can specify and efficiently monitor complex properties for hierarchical structures over infinite alphabets. Streaming tree transducers can express and efficiently process complex XML transformations while enjoying decidable procedures
1 edition published in 2015 in English and held by 1 WorldCat member library worldwide
First, we introduce succinct models of transducers that can operate over large alphabets and design BEX, a language for analysing string coders. We use BEX to prove the correctness of UTF and B ASE64 encoders and decoders. Next, we develop a theory of tree transducers over infinite alphabets and design FAST, a language for analysing treemanipulating programs. We use FAST to detect vulnerabilities in HTML sanitizers, check whether augmented reality taggers conflict, and optimize and analyze functional programs that operate over lists and trees. Finally, we focus on laying the foundations of stream processing of hierarchical data such as XML files and program traces. We introduce two new efficient and executable models that can process the input in a lefttoright linear pass: symbolic visibly pushdown automata and streaming tree transducers. Symbolic visibly pushdown automata are closed under Boolean operations and can specify and efficiently monitor complex properties for hierarchical structures over infinite alphabets. Streaming tree transducers can express and efficiently process complex XML transformations while enjoying decidable procedures
A robust model for finding optimal evolutionary trees(
Book
)
1 edition published in 1993 in English and held by 1 WorldCat member library worldwide
Constructing evolutionary trees for species sets is a fundamental problem in biology. One of the standard models assumes the ability to compute distances between every pair of species ... The primary contributions of this paper are as follows: (1) We suggest a model of computation and efficient algorithms for constructing optimal ultrametric trees under this model. This is the first reasonable model for which optimal trees can be found in polynomial time. (2) We show that for another criterion, there is an [epsilon] [greater than] 0 such that unless P = NP, finding the optimal ultrametric tree can not be approximated
1 edition published in 1993 in English and held by 1 WorldCat member library worldwide
Constructing evolutionary trees for species sets is a fundamental problem in biology. One of the standard models assumes the ability to compute distances between every pair of species ... The primary contributions of this paper are as follows: (1) We suggest a model of computation and efficient algorithms for constructing optimal ultrametric trees under this model. This is the first reasonable model for which optimal trees can be found in polynomial time. (2) We show that for another criterion, there is an [epsilon] [greater than] 0 such that unless P = NP, finding the optimal ultrametric tree can not be approximated
Detecting Wikipedia Vandalism via SpatioTemporal Analysis of Revision Metadata(
)
1 edition published in 2010 in English and held by 0 WorldCat member libraries worldwide
Blatantly unproductive edits undermine the quality of the collaborativelyedited encyclopedia, Wikipedia. They not only disseminate dishonest and offensive content, but force editors to waste time undoing such acts of vandalism. Language processing has been applied to combat these malicious edits, but as with email spam, these filters are evadable and computationally complex. Meanwhile, recent research has shown spatial and temporal features effective in mitigating email spam, while being lightweight and robust. In this paper, we leverage the spatiotemporal properties of revision metadata to detect vandalism on Wikipedia. An administrative form of reversion called rollback enables the tagging of malicious edits, which are contrasted with nonoffending edits in numerous dimensions. Crucially, none of these features require inspection of the article or revision text. Ultimately, a classifier is produced which flags vandalism at performance comparable to the naturallanguage efforts we intend to complement (85% accuracy at 50% recall). The classifier is scalable (processing 100+ edits a second) and has been used to locate over 5,000 manuallyconfirmed incidents of vandalism outside our labeled set
1 edition published in 2010 in English and held by 0 WorldCat member libraries worldwide
Blatantly unproductive edits undermine the quality of the collaborativelyedited encyclopedia, Wikipedia. They not only disseminate dishonest and offensive content, but force editors to waste time undoing such acts of vandalism. Language processing has been applied to combat these malicious edits, but as with email spam, these filters are evadable and computationally complex. Meanwhile, recent research has shown spatial and temporal features effective in mitigating email spam, while being lightweight and robust. In this paper, we leverage the spatiotemporal properties of revision metadata to detect vandalism on Wikipedia. An administrative form of reversion called rollback enables the tagging of malicious edits, which are contrasted with nonoffending edits in numerous dimensions. Crucially, none of these features require inspection of the article or revision text. Ultimately, a classifier is produced which flags vandalism at performance comparable to the naturallanguage efforts we intend to complement (85% accuracy at 50% recall). The classifier is scalable (processing 100+ edits a second) and has been used to locate over 5,000 manuallyconfirmed incidents of vandalism outside our labeled set
Computing Diameter in the Streaming and SlidingWindow Models (Preprint)(
)
1 edition published in 2002 in English and held by 0 WorldCat member libraries worldwide
We investigate the diameter problem in the streaming and slidingwindow models. We show that, for a stream of n points or a sliding window of size n, any exact algorithm for diameter requires Omega(n) bits of space. We present a simple epsilonapproximation algorithm for computing the diameter in the streaming model. Our main result is an epsilonapproximation algorithm that maintains the diameter in two dimensions in the sliding windows model using Omicron [(1/epsilon (exp 3/2) log3 n(log R + log log n + log 1/epsilon))] bits of space, where R is the maximum, over all windows, of the ratio of the diameter to the minimum nonzero distance between any two points in the window
1 edition published in 2002 in English and held by 0 WorldCat member libraries worldwide
We investigate the diameter problem in the streaming and slidingwindow models. We show that, for a stream of n points or a sliding window of size n, any exact algorithm for diameter requires Omega(n) bits of space. We present a simple epsilonapproximation algorithm for computing the diameter in the streaming model. Our main result is an epsilonapproximation algorithm that maintains the diameter in two dimensions in the sliding windows model using Omicron [(1/epsilon (exp 3/2) log3 n(log R + log log n + log 1/epsilon))] bits of space, where R is the maximum, over all windows, of the ratio of the diameter to the minimum nonzero distance between any two points in the window
ASCRED: Reputation Service for Trustworthy Interdomain Routing(
)
1 edition published in 2010 in English and held by 0 WorldCat member libraries worldwide
The current design of BGP implicitly assumes the existence of trust between ASes with respect to exchanging valid BGP updates. This assumption of complete trust is problematic given the frequent announcement of invalid  inaccurate or unnecessary  updates. This paper presents ASCRED, a reputation service for ASes which quantifies the level of trust one can have with respect to its announcing valid updates. To compute the reputation, ASCRED analyzes the past updates announced by each observable AS in the Internet, over a timewindow, based on welldefined properties. It then classifies the resulting observations into multiple types of feedback. The feedback values are input into a mathematical function for computing AS reputation. The reputation is then used to track the instances of invalid updates announced in the Internet and trigger alerts. The contributions of the paper are: (1) a reputation service for ASes, characterizing their trustworthiness; (2) a set of well defined properties for analyzing AS behavior; (3) a simple reputation function and feedback mechanism; (4) a reputation portal which regularly publishes AS reputation; and (5) a reputationbased alert service which tracks potentially invalid updates in the Internet. Detailed analysis of ASCRED demonstrates: (1) AS behavior is repetitive making reputation an effective trust metric, and (2) ASCRED's alerts for invalid updates show an eight fold improvement over existing alert systems
1 edition published in 2010 in English and held by 0 WorldCat member libraries worldwide
The current design of BGP implicitly assumes the existence of trust between ASes with respect to exchanging valid BGP updates. This assumption of complete trust is problematic given the frequent announcement of invalid  inaccurate or unnecessary  updates. This paper presents ASCRED, a reputation service for ASes which quantifies the level of trust one can have with respect to its announcing valid updates. To compute the reputation, ASCRED analyzes the past updates announced by each observable AS in the Internet, over a timewindow, based on welldefined properties. It then classifies the resulting observations into multiple types of feedback. The feedback values are input into a mathematical function for computing AS reputation. The reputation is then used to track the instances of invalid updates announced in the Internet and trigger alerts. The contributions of the paper are: (1) a reputation service for ASes, characterizing their trustworthiness; (2) a set of well defined properties for analyzing AS behavior; (3) a simple reputation function and feedback mechanism; (4) a reputation portal which regularly publishes AS reputation; and (5) a reputationbased alert service which tracks potentially invalid updates in the Internet. Detailed analysis of ASCRED demonstrates: (1) AS behavior is repetitive making reputation an effective trust metric, and (2) ASCRED's alerts for invalid updates show an eight fold improvement over existing alert systems
more
fewer
Audience Level
0 

1  
Kids  General  Special 
Related Identities
 Naor, Moni
 Rudich, Steven
 Warnow, Tandy
 Santha, Miklos
 University of Pennsylvania Department of Computer and Information Science
 Lee, Insup
 West, Andrew G.
 PENNSYLVANIA UNIV PHILADELPHIA Dept. of COMPUTER AND INFORMATION SCIENCE
 Yao, Andrew ChiChih
 University of Warwick Department of Computer Science
Languages