Search 250 million research papers

String kernel

SpringerReference · 2012

Cited by 0

Bootstrapping the String KLT Kernel

Alan Shih-Kuan Chen, Henriette Elvang, Aidan Herderschee · arXiv (Cornell University) · 2023

We show that a generalized version of the 4-point string theory KLT double-copy map is the most general solution to the minimal-rank double-copy bootstrap in effective field theory. This follows from significant restrictions of the 4-point map resulting from the 6-point bootstrap analysis. The generalized 4-point double-copy map is defined by a function with only two parameters times a simple function that is symmetric in $s,t,u$. The two parameters can be interpreted as independent choices of $α'$, one for each of the two sets of amplitudes double-copied with the map. Specifically, each of those two sets of amplitudes must obey either the string monodromy relations or the field theory KK & BCJ relations; there are no other options. We propose a closed form of the new double-copy map that interpolates between the original KLT string double-copy and the open & closed string period integrals. The construction clarifies the "single-valued projection" property of the Riemann zeta-function values for the 4-point string theory double copy.

Cited by 0Open access

Sketch recognition via string kernel

Shizhong Liao, Menghua Duan · 2012

Sketch recognition is one of the essential step of sketch understanding. Challenge in sketch recognition is the variation and imprecision present in sketch. Free drawing styles of sketching make it difficult to build a robust sketch recognition system. This paper proposes a novel recognition approach that can recognize primitive shapes, as well as combinations of these primitives. The approach is independent of stroke order, number, as well as invariant to size and aspect ratio of sketch. Feature string is used to represent primitives. We defined a similarity measure on these feature strings that counts common substrings in two input strings, which is referred to as the string kernel in the field of kernel methods. Support vector machine(SVM) is then trained with labeled examples to handle the task of classification. The experiment on hand drawn digital circuit diagrams shows that our system can recognize sketching efficiently and robustly.

Cited by 5

Meta learning on string kernel SVMs for string categorization

Nuwan Gunasekara · AUT Scholarly Commons · 2010

Even though Support Vector Machines (SVMs) are capable of identifying patterns in high dimensional spaces that are presented by kernels without a computational decrease, their performance is determined by two main factors: SVM cost param- eter and kernel parameters. It is empirically proven that the optimum parameter combination for both SVM cost and kernel yield high pattern recognition accuracies in SVMs. Various methods are discussed in the literature to optimize these SVM parameters, namely trial and error method, grid optimization, leave-one-out cross validation, generalization error estimation using gradient descent, and evolutionary algorithms. However, these optimization methods have downfalls: the trial and error method is considered to be imprecise and unreliable (Zhang & He, 2010), the grid parameter search requires complex computation (Friedrichs & Igel, 2005), leave-one- out cross validation is considered to be computationally expensive, and the gradient descent method can fall into a local `minima' (Zhang & He, 2010). Also, these meth- ods mainly optimize numeric kernels. However, when optimizing string kernel SVMs in string classi¯cation, researchers are confronted with two main obstacles. Firstly, there is little or no literature pertaining to string kernel SVM optimization at the moment. Secondly, according to the experiments: Rieck and Laskov (2007), Lodhi, Saunders, Shawe-Taylor, Cristianini, and Watkins (2002), Sharma, Girolami, and Sventek (2007) and S.Sonnenburg (2006), string dataset characteristics in°uence the optimum parameter combination of string kernel SVMs. The current string kernel SVM optimization methods do not take string dataset characteristics into account. Baring this in mind, the thesis initially attempts to identify a mechanism to extract string dataset characteristics from a string dataset. It then, attempts to derive a string kernel SVM optimization method which uses these extracted string dataset characteristics. The initial objective is achieved by explaining that a sting dataset is presentable in token-frequency space (where each dimension is represented by a token) and then computing string meta-features by the points of this token-frequency space. Now for a machine learning algorithm, a meta model is trained using computed string meta-features for each dataset in a string dataset pool, learning algorithm parameters, and accuracy information. This trained meta model helps to predict string classi¯cation accuracy for a new string dataset by computing relevant string meta-features. This principle is employed to optimize string kernel SVMs in the proposed meta learning algorithm (this ful¯lls the second objective of the thesis) where it only computes the relevant string meta-features to yield optimum parameter combination for the machine learning algorithm on the new string dataset. In the experiments, three string kernel SVMs, edit-distance SVM, bag-of-words SVM and n-gram SVM, were optimized using the proposed algorithm on four string datasets: spam, Reuters-21578, Network Application Detection and e-News Cate- gorization. The experiment results revealed that the algorithm was able to produce parameter combinations which yield good string classi¯cation accuracies for string kernel SVMs on most string datasets. It is also revealed that some string kernel SVMs may not be suitable on certain string datasets. As future work, one could research for more string meta-features that help to employ meta learning on string classi¯cation more e®ectively. Also, one could introduce upper and lower bounds to the proposed algorithm, which help the predicted string classi¯cation accuracy to be within the range (0,100). Further experiments on more string datasets will help to identify the robustness of the algorithm.

Cited by 1Open access

Syllables and Other String Kernel extensions

Tschach, Hauke, Saunders, Craig, Shawe-Taylor, John · ePrints Soton (University of Southampton) · 2002

During the last years, the use of string kernels that compare documents has been shown to achieve good results on text classification problems. In this paper we introduce the application of the string kernel in conjunction with syllables. Using syllables shortens the representation of documents compared to a character based representation and as a result reduces computation time. Moreover syllables provide a more natural representation of text; rather than the traditional coarse representation given by the bag-of-words, or the too fine one resulting from considering individual letters only. We give some experimental results which show that syllables can be effectively used in text-categorisation problems. In this paper we also propose two extensions to the string kernel. The first introduces a lambda-weighting scheme, where different symbols can be given differing decay weightings. This may be useful in text and other applications where the insertion of certain symbols may be known to be less significant. We also introduce the concept of ‘soft matching’, where symbols can match (possibly weighted by relevance) even if they are not identical. Again, this provides a method of incorporating prior knowledge where certain symbols can be regarded as a partial or exact match and contribute to the overal similarity measure for two data items. We also give an overview of other kernels for comparing strings that have evolved over the past few years. In the appendix we give a detailed description of a method for computing the string kernel efficiently.

Cited by 2Open access

Efficient Global String Kernel with Random Features

Lingfei Wu, Ian En-Hsu Yen, Siyu Huo et al. · 2019

Analysis of large-scale sequential data has been one of the most crucial tasks in areas such as bioinformatics, text, and audio mining. Existing string kernels, however, either (i) rely on local features of short substructures in the string, which hardly capture long discriminative patterns, (ii) sum over too many substructures, such as all possible subsequences, which leads to diagonal dominance of the kernel matrix, or (iii) rely on non-positive-definite similarity measures derived from the edit distance. Furthermore, while there have been works addressing the computational challenge with respect to the length of string, most of them still experience quadratic complexity in terms of the number of training samples when used in a kernel-based classifier. In this paper, we present a new class of global string kernels that aims to (i) discover global properties hidden in the strings through global alignments, (ii) maintain positive-definiteness of the kernel, without introducing a diagonal dominant kernel matrix, and (iii) have a training cost linear with respect to not only the length of the string but also the number of training string samples. To this end, the proposed kernels are explicitly defined through a series of different random feature maps, each corresponding to a distribution of random strings. We show that kernels defined this way are always positive-definite, and exhibit computational benefits as they always produce Random String Embeddings (RSE) that can be directly used in any linear classification models. Our extensive experiments on nine benchmark datasets corroborate that RSE achieves better or comparable accuracy in comparison to state-of-the-art baselines, especially with the strings of longer lengths. In addition, we empirically show that RSE scales linearly with the increase of the number and the length of string.

Cited by 1

Inverse of the String Theory KLT Kernel

Mizera, Sebastian · arXiv (Cornell University) · 2016

The field theory Kawai-Lewellen-Tye (KLT) kernel, which relates scattering amplitudes of gravitons and gluons, turns out to be the inverse of a matrix whose components are bi-adjoint scalar partial amplitudes. In this note we propose an analogous construction for the string theory KLT kernel. We present simple diagrammatic rules for the computation of the $α'$-corrected bi-adjoint scalar amplitudes that are exact in $α'$. We find compact expressions in terms of graphs, where the standard Feynman propagators $1/p^2$ are replaced by either $1/\sin (πα' p^2/2)$ or $1/\tan (πα' p^2/2)$, as determined by a recursive procedure. We demonstrate how the same object can be used to conveniently expand open string partial amplitudes in a BCJ basis.

Cited by 0Open access

A Unified String Kernel for Biology Sequence

Yuan De-hui, Shengyun Yang, Guoming Lai · Lecture notes in computer science · 2008

Cited by 0

Bootstrapping the String Kawai-Lewellen-Tye Kernel

Alan Shih-Kuan Chen, Henriette Elvang, Aidan Herderschee · Physical Review Letters · 2023

We show that double-copy maps for amplitudes in effective field theory are severely constrained at four points by self-consistency and locality at six points. The resulting double-copy kernel depends only on two parameters as well as a specific symmetric function in s, t, u and interpolates between the original Kawai-Lewellen-Tye (KLT) string double copy and the open and closed string period integrals. Amplitudes double copied with this map must obey either the string monodromy relations or the field theory Kleiss-Kuijf (KK) and Bern, Carrasco, and Johansson (BCJ) relations; there are no other options. Our construction elucidates the "single-valued projection" property of the Riemann zeta-function values for the four-point string theory double copy.

Cited by 16Open access

A weighted string kernel for protein fold recognition

Saghi Nojoomi, Koehl, Patrice · Figshare · 2017

Abstract Background Alignment-free methods for comparing protein sequences have proved to be viable alternatives to approaches that first rely on an alignment of the sequences to be compared. Much work however need to be done before those methods provide reliable fold recognition for proteins whose sequences share little similarity. We have recently proposed an alignment-free method based on the concept of string kernels, SeqKernel (Nojoomi and Koehl, BMC Bioinformatics, 2017, 18:137). In this previous study, we have shown that while Seqkernel performs better than standard alignment-based methods, its applications are potentially limited, because of biases due mostly to sequence length effects. Methods In this study, we propose improvements to SeqKernel that follows two directions. First, we developed a weighted version of the kernel, WSeqKernel. Second, we expand the concept of string kernels into a novel framework for deriving information on amino acids from protein sequences. Results Using a dataset that only contains remote homologs, we have shown that WSeqKernel performs remarkably well in fold recognition experiments. We have shown that with the appropriate weighting scheme, we can remove the length effects on the kernel values. WSeqKernel, just like any alignment-based sequence comparison method, depends on a substitution matrix. We have shown that this matrix can be optimized so that sequence similarity scores correlate well with structure similarity scores. Starting from no information on amino acid similarity, we have shown that we can derive a scoring matrix that echoes the physico-chemical properties of amino acids. Conclusion We have made progress in characterizing and parametrizing string kernels as alignment-based methods for comparing protein sequences, and we have shown that they provide a framework for extracting sequence information from structure.

Cited by 0Open access

String Kernel-Based Techniques for Native Language Identification

Vamshi Kumar Gurram, Sanil Jayamohan, V. S. Anoop et al. · Human-Centric Intelligent Systems · 2023

Abstract In recent years, Native Language Identification (NLI) has shown significant interest in computational linguistics. NLI uses an author’s speech or writing in a second language to figure out their native language. This may find applications in forensic linguistics, language teaching, second language acquisition, authorship attribution, identification of spam emails or phishing websites, etc. Conventional pairwise string comparison techniques are computationally expensive and time-consuming. This paper presents fast NLI techniques based on string kernels such as spectrum, presence bits, and intersection string kernels incorporating different learners such as a Support Vector Machine (SVM), Random Forest (RF), and Extreme Gradient Boosting-XGBoost (XGB). Feature sets for the proposed techniques are generated using different combinations of features such as n-word grams and noun phrases. Experimental analyses are carried out using 8235 English as a second language articles from 10 different linguistic backgrounds from a typical NLP benchmark dataset. The experimental results show that the proposed NLI technique incorporating a spectrum string kernel with an RF classifier outperformed existing character n-gram string kernels incorporating SVM, RF, and XGB classifiers. Also, comparable results were observed among different combinations of string kernels. Interestingly, the random forest classifier outperformed SVM and XGB classifiers with different feature sets. All the proposed NLI techniques demonstrated promising results with significant improvement in training time, with the best result attaining more than a 95 percent decrease in training time. The reduced training time of proposed techniques makes it well suited to scale NLI applications for production.

Cited by 4Open access

A weighted string kernel for protein fold recognition

Saghi Nojoomi, Koehl, Patrice · Figshare · 2017

Abstract Background Alignment-free methods for comparing protein sequences have proved to be viable alternatives to approaches that first rely on an alignment of the sequences to be compared. Much work however need to be done before those methods provide reliable fold recognition for proteins whose sequences share little similarity. We have recently proposed an alignment-free method based on the concept of string kernels, SeqKernel (Nojoomi and Koehl, BMC Bioinformatics, 2017, 18:137). In this previous study, we have shown that while Seqkernel performs better than standard alignment-based methods, its applications are potentially limited, because of biases due mostly to sequence length effects. Methods In this study, we propose improvements to SeqKernel that follows two directions. First, we developed a weighted version of the kernel, WSeqKernel. Second, we expand the concept of string kernels into a novel framework for deriving information on amino acids from protein sequences. Results Using a dataset that only contains remote homologs, we have shown that WSeqKernel performs remarkably well in fold recognition experiments. We have shown that with the appropriate weighting scheme, we can remove the length effects on the kernel values. WSeqKernel, just like any alignment-based sequence comparison method, depends on a substitution matrix. We have shown that this matrix can be optimized so that sequence similarity scores correlate well with structure similarity scores. Starting from no information on amino acid similarity, we have shown that we can derive a scoring matrix that echoes the physico-chemical properties of amino acids. Conclusion We have made progress in characterizing and parametrizing string kernels as alignment-based methods for comparing protein sequences, and we have shown that they provide a framework for extracting sequence information from structure.

Cited by 0Open access

A weighted string kernel for protein fold recognition

Saghi Nojoomi, Patrice Koehl · BMC Bioinformatics · 2017

We have made progress in characterizing and parametrizing string kernels as alignment-based methods for comparing protein sequences, and we have shown that they provide a framework for extracting sequence information from structure.

Cited by 3Open access

Entangled kernel of coded information using quantum string

Abdulla Eid, N. Metwally · International Journal of Quantum Information · 2023

The possibility of encoding classical string of information into a quantum string via CNOT operation and Hadamard gate is discussed. It has been shown that maximizing/minimizing the quantum correlation of the generated quantum string depends on the values of the initial classical strings. The kernel of each string, as well as, the total string had been examined, where by applying the Hadamard gate on the classical string, one can decrease its similarity as well as the similarity of the total kernel. However, in the absence of the Hadamard gate, the large similarity of the final quantum string has been observed. The kernel of the final quantum string may be used as an indictor of the security degree of the encoded information, where the large degree of security has depicted at small kernel. The high degree of security can be implemented by applying the Hadamard gate on only one classical string.

Cited by 2

A multi-fold string kernel for sequence classification

Aniruddha Maiti, Santanu Ghorai, Anirban Mukherjee · 2015

A novel framework is proposed to classify biological sequences using a kernel. It considers the topological information along with the primary structural information. The widely used string kernel for sequence classification does not take into account the structural information which might be available for biological sequences. The proposed kernels incorporate the additional structural information and thus make the features more informative.

Cited by 0

R.: Genre classification via an lz78-based string kernel

Ming Li, Ronan Sleep · 2005

We develop the notion of normalized information distance (NID) [7] into a kernel distance suitable for use with a Support Vector Machine classifier, and demonstrate its use for an audio genre classification task. Our classification scheme involves a relatively small number of low-level audio features, is efficient to compute, yet generates an accuracy which compares well with recent works.

Cited by 0

String Kernel Approach for Efficient Extraction of Medical Relations

V Vrinda., Rajasekar Venkatesan · International Journal of Computer Applications · 2015

This paper presents a methodology for building an application that can classify healthcare information. It extracts informative sentences from medical papers that mentions about diseases and treatments, and then recognizes semantic relations that exist between the entities in the informative sentences. Support Vector Machine algorithm with String Kernel is used for relation identification. The proposed system avoids unnecessary information and gives the user disease and Treatment related sentences from medical pages. Evaluation results for this approach show that the proposed methodology obtains reliable results. The string kernel approach is also compared with nave bayes approach and it was found that string kernel approach outperforms nave bayes method for classification. This technique can be integrated with any medical management system to make good decisions and in patient management system by automatically mining the biomedical information from digital repositories. This system enables easy access to medical information in rural areas where there is a relative shortage of physicians.

Cited by 0Open access

siRNA silencing efficacy prediction using the RNA string kernel

Shibin Qiu, Terran Lane · International Journal of Computational Biology and Drug Design · 2008

While most existing string kernels are developed for general purpose sequences and have been applied to text and protein classifications, the RNA string kernel is particularly designed to model mismatches, G-U wobbles, and bulges of RNA biology. We adapt the RNA kernel to compute the similarity of the short interfering RNAs (siRNAs), initiators of RNA interference, and use it in support vector regression to predict the siRNA silencing efficacy treated as a continuous variable. Empirical results on biological data sets demonstrate that the RNA string kernel performed favourably. In addition, it is simple to implement and fast to compute.

Cited by 0

Increasing Classification Accuracy in libSVM Using String Kernel Functions

Homoliak, Ivan · Brno University of Technology Digital Library (Brno University of Technology) · 2010

Práce se zabýva zkoumáním závislostí kvality klasifikace textových řetězců na vlastnostech vybraných řetězcových funkcí, použitých na určení míry podobnosti dvou textových řetezců. Práce přešetřuje též kombinování výsledků řetězcových funkcí aritmetickými operacemi plus a krát. Získané výsledky se v práci aplikují na detekci nevyžádané elektronické pošty.

Cited by 0

Efficient geometric-based computation of the string subsequence kernel

Slimane Bellaouar, Hadda Cherroun, Djelloul Ziadi · Data Mining and Knowledge Discovery · 2017

Cited by 0Open access