11 The Geometric Basis of Semi-supervised Learning (2008)
Vikas Sindhwani, Misha Belkin, Partha Niyogi
In this chapter, we present an algorithmic framework for semi-supervised inference based on geometric properties of probability distributions. Our approach brings together Laplacian-based spectral...
A Regularity conditions on p and S We make the following assumptions about p: (2008)
Hariharan Narayanan, Mikhail Belkin, Partha Niyogi
1. p can be extended to a function p ′ that is L−Lipshitz and which is bounded above by pmax. 2. For 0 < t < t0, min(p(x), Kt(x, y)p(y)dy) ≥ pmin. Note that this is a property of both of...
Abstract Towards a Theoretical Foundation for Laplacian-Based Manifold Methods (2008)
In recent years manifold methods have attracted a considerable amount of attention in machine learning. However most algorithms in that class may be termed “manifold-motivated ” as they lack any...
Mikhail Belkin, Hariharan Narayanan, Partha Niyogi
flow and a faster algorithm to compute the
Mercer’s Theorem, Feature Maps, and Smoothing (2008)
Ha Quang Minh, Partha Niyogi, Yuan Yao
Abstract. We study Mercer’s theorem and feature maps for several positive definite kernels that are widely used in practice. The smoothing properties of these kernels will also be explored. 1
Partha Niyogi, Stephen Smale, Shmuel Weinberger
Abstract. Recently there has been a lot of interest in geometrically motivated approaches to data analysis in high-dimensional spaces. We consider the case where data is drawn from sampling a...
Competing Models of Linguistic Change: Evolution and Beyond. In commemoration of Eugenio (2008)
Dinoj Surendran, Partha Niyogi, Coseriu Amsterdam, Philadelphia Benjamins
Quantifying the functional load of phonemic oppositions, distinctive features, and suprasegmentals
Many problems in information processing involve some form of dimensionality reduction. In this paper, we introduce Locality Preserving Projections (LPP). These are linear projective maps that arise...
Epsilon Focusing---A Strategy For Active Example Selection (2007)
Partha Niyogi Artificial, Partha Niyogi
this paper, we would like to explore the possibility of a more active scheme where the learner collects examples that are relevant to the task at hand.
Epsilon Focusing---A Strategy for Active Example Selection (2007)
Partha Niyogi Lucent, Partha Niyogi
In most classical formulations of learning from examples, a passive learner is presented with examples randomly drawn. Here we discuss an ffl-focusing strategy that actively choose examples for...
BACKGROUND MOTIVATION: TRIGGERS AND LANGUAGE ACQUISITION (2007)
Several Researchers, Including Gibson, Henceforth Gw, Partha Niyogi, Robert C. Berwick
This paper shows how to formally characterize lan-guage learning in a finite parameter space as a Markov structure, hnportant new language learning results fol-low directly: explicitly calculated...
Modelling Speaker Variability and hnposing Speaker Constraints in Phonetic Classificat, ion (2007)
Partha Niyogi, Partha Niyogi, Arthur C. Smith
by
C.B.C.L. Paper No. 115 The Logical Problem of Language Change (2007)
Partha Niyogi, Robert C. Berwick
This publication can be retrieved by anonymous ftp to publications.ai.mit.edu. This paper considers the problem of language change. Linguists must explain not only how languages are learned but also...
This publication can be retrieved by anonymous ftp to publications.ai.mit.edu. In Phys. Rev. Letters, 73:2, 5 Dec. 94, Mantegna et al. conclude on the basis of Zipf rank frequency data that noncoding...
The interaction of stability and weakness in (2007)
We provide an analysis of AdaBoost within the framework of algorithmic stability. In particular, we show that AdaBoost is a stabilitypreserving operation: if the \input " (the weak learner)...
The interaction of stability and weakness in (2007)
We provide an analysis of AdaBoost within the framework of algorithmic stability. In particular, we show that AdaBoost is a stabilitypreserving operation: if the “input ” (the weak learner) to...
Mukherjee, Sayan, Niyogi, Partha, Poggio, Tomaso, Rifkin, Ryan
Solutions of learning problems by Empirical Risk Minimization (ERM) -- and almost-ERM when the minimizer does not exist -- need to be consistent, so that they may be predictive. They also need to be...
Niyogi, Partha, Girosi, Federico
In this paper, we bound the generalization error of a class of Radial Basis Function networks, for certain well defined function learning tasks, in terms of the number of parameters and number of...
Mikhail Belkin, Partha Niyogi, Vikas Sindhwani, Peter Bartlett
We propose a family of learning algorithms based on a new form of regularization that allows us to exploit the geometry of the marginal distribution. We focus on a semi-supervised framework that...
On the relation between low density separation, spectral clustering and graph cuts (2006)
Hariharan Narayanan, Mikhail Belkin, Partha Niyogi
One of the intuitions underlying many graph-based methods for clustering and semi-supervised learning, is that class or cluster boundaries pass through areas of low probability density. In this paper...
On the relation between low density separation, spectral clustering and graph cuts (2006)
Hariharan Narayanan, Mikhail Belkin, Partha Niyogi
One of the intuitions underlying many graph-based methods for clustering and semi-supervised learning, is that class or cluster boundaries pass through areas of low probability density. In this paper...
Regularization Approaches in Learning Theory (2006)
Lorenzo Rosasco, Lorenzo Rosasco, Ernesto De Vito, Partha Niyogi
2 Learning from examples can be seen as a very general framework for modeling a variety of different statistical inference problems. Such statistical problems are at the basis of the design of...
A co-regularized approach to semi-supervised learning with multiple views (2005)
Vikas Sindhwani, Partha Niyogi
The Co-Training algorithm uses unlabeled examples in multiple views to bootstrap classifiers in each view, typically in a greedy manner, and operating under assumptions of view-independence and...
Face recognition using laplacianfaces (2005)
Xiaofei He, Shuicheng Yan, Yuxiao Hu, Partha Niyogi, Hong-jiang Zhang
Abstract—We propose an appearance-based face recognition method called the Laplacianface approach. By using Locality Preserving Projections (LPP), the face images are mapped into a face subspace...
Face recognition using laplacianfaces (2005)
Xiaofei He, Shuicheng Yan, Yuxiao Hu, Partha Niyogi, Hong-jiang Zhang
Abstract—We propose an appearance-based face recognition method called the Laplacianface approach. By using Locality Preserving Projections (LPP), the face images are mapped into a face subspace...
Towards a theoretical foundation for Laplacian-based manifold methods (2005)
Abstract. In recent years manifold methods have attracted a considerable amount of attention in machine learning. However most algorithms in that class may be termed “manifold-motivated ” as they...
Linear manifold regularization for large scale semi-supervised learning (2005)
Vikas Sindhwani, Partha Niyogi
The enormous wealth of unlabeled data in many applications of machine learning is beginning to pose challenges to the designers of semi-supervised learning methods. We are interested in developing...
Robust Acoustic Object Detection (2005)
Yali Amit, Alexey Koloydenko, Partha Niyogi
We consider a novel approach to the problem of detecting phonological objects like phonemes, syllables, or words, directly from the speech signal. We begin by dening local features in the...
On Manifold Regularization (2005)
Mikhail Belkin, Partha Niyogi, Vikas Sindhwani
We propose a family of learning algorithms based on a new form of regularization that allows us to exploit the geometry of the marginal distribution. We focus on a semisupervised framework that...
Laplacian score for feature selection (2005)
Xiaofei He, Deng Cai, Partha Niyogi
In supervised learning scenarios, feature selection has been studied widely in the literature. Selecting features in unsupervised learning scenarios is a much harder problem, due to the absence of...
Tensor subspace analysis (2005)
Xiaofei He, Deng Cai, Partha Niyogi
Previous work has demonstrated that the image variations of many objects (human faces in particular) under variable lighting can be effectively modeled by low dimensional linear spaces. The typical...
Beyond the point cloud: from transductive to semi-supervised learning (2005)
Vikas Sindhwani, Partha Niyogi
Due to its occurrence in engineering domains and implications for natural learning, the problem of utilizing unlabeled data is attracting increasing attention in machine learning. A large body of...
Stability and generalization of bipartite ranking algorithms (2005)
Shivani Agarwal, Partha Niyogi
Abstract. The problem of ranking, in which the goal is to learn a real-valued ranking function that induces a ranking or ordering over an instance space, has recently gained attention in machine...
Face recognition using laplacianfaces (2005)
Xiaofei He, Shuicheng Yan, Yuxiao Hu, Partha Niyogi, Hong-jiang Zhang
We propose an appearance based face recognition method called the Laplacianface approach. By using Locality Preserving Projections (LPP), the face images are mapped into a face subspace for analysis....
Feature selection in MLPs and SVMs based on maximum output information (2004)
Vikas Sindhwani, Subrata Rakshit, Dipti Deodhare, Deniz Erdogmus, Jose Principe, Partha Niyogi
Abstract — We present feature selection algorithms for multilayer
Regularization and semi-supervised learning on large graphs (2004)
Mikhail Belkin, Irina Matveeva, Partha Niyogi
Abstract. We consider the problem of labeling a partially labeled graph. This setting may arise in a number of situations from survey sampling to information retrieval to pattern recognition in...
Manifold regularization: A geometric framework for learning from examples (2004)
Mikhail Belkin, Partha Niyogi, Vikas Sindhwani, Peter Bartlett
We propose a family of learning algorithms based on a new form of regularization that allows us to exploit the geometry of the marginal distribution. We focus on a semi-supervised framework that...
Optimizing the mutual intelligibility of linguistic agents in a shared world (2004)
Natalia Komarova, Partha Niyogi
Abstract: We consider the problem of linguistic agents that communicate with each other about a shared world. We develop a formal notion of a language as a set of probabilistic associations between...
Semi-supervised learning on Riemannian manifolds (2004)
We consider the general problem of utilizing both labeled and unlabeled data to improve classification accuracy. Under the assumption that the data lie on a submanifold in a high dimensional space,...
Regularization and semi-supervised learning on large graphs (2004)
Mikhail Belkin, Irina Matveeva, Partha Niyogi
Abstract. We consider the problem of labeling a partially labeled graph. This setting may arise in a number of situations from survey sampling to information retrieval to pattern recognition in...
On Manifold Regularization Mikhail Belkin (2004)
Partha Niyogi, Vikas Sindhwani
We propose a family of learning algorithms based on a new form of regularization which allows us to incorporate both labeled and unlabeled data in a general-purpose learner. Transductive graph...
Semi-supervised learning on Riemannian manifolds (2004)
We consider the general problem of utilizing both labeled and unlabeled data to improve classification accuracy. Under the assumption that the data lie on a submanifold in a high dimensional space,...
Measuring the Functional Load of Phonological Contrasts (2003)
Surendran, Dinoj, Niyogi, Partha
Frequency counts are a measure of how much use a language makes of a linguistic unit, such as a phoneme or word. However, what is often important is not the units themselves, but the contrasts...
Laplacian Eigenmaps for Dimensionality Reduction and Data Representation (2003)
Abstract One of the central problems in machine learning and pattern recognition is to develop appropriate representations for complex data. We consider the problem of constructing a representation...
Ankan Saha, Advisor Prof, Partha Niyogi
Using Conditional Random Fields(CRF) as a framework for probabilistic segmentation and labeling of sequenced data, we handle the traditional problem of POS tagging using a modular structure in which...
Measuring the Usefulness (Functional Load) of Phonological Contrasts (2003)
Dinoj Surendran, Partha Niyogi
This paper describes a method to measure how much use a language makes of a contrast to convey information, i.e. the functional load (FL) of the contrast
Sayan Mukherjee, Partha Niyogi, Tomaso Poggio, Ryan Rifkin
Solutions of learning problems by Empirical Risk Minimization (ERM) -- and almost-ERM when the minimizer does not exist -- need to be consistent, so that they may be predictive. They also need to be...
Locality Preserving Projections (2003)
Many problems in information processing involve some form of dimensionality reduction. In this paper, we introduce Locality Preserving Projections (LPP). These are linear projective maps that arise...
Measuring the Functional Load of Phonological Contrasts (2003)
Dinoj Surendran, Partha Niyogi
This paper describes a method to measure how much use a language makes of a contrast to convey information, i.e. the functional load (FL) of the contrast
Mukherjee, Sayan, Niyogi, Partha, Poggio, Tomaso, Rifkin, Ryan
revised July 2003
Mukherjee, Sayan, Niyogi, Partha, Poggio, Tomaso, Rifkin, Ryan
revised July 2003
generalization and necessary and sufficient for consistency of Empirical Risk Minimization (2002)
Sayan Mukherjee, Partha Niyogi, Tomaso Poggio, Ryan Rifkin, Sayan Mukherjee, Partha Niyogi, ...
Solutions of learning problems by Empirical Risk Minimization (ERM) – and almost-ERM when the minimizer does not exist – need to be consistent, so that they may be predictive. They also need to...
Detecting and interpreting acoustic features by support vector machines (Tech (2002)
1 INTRODUCTION Any approach to speech perception or recognition will have to specify a mechanism by means of which the acoustic input is mapped to discrete linguistic objects or symbols. In most...
The Computational Study of Diachronic Linguistics (2002)
At the heart of these models is the subtle interplay between language
Laplacian eigenmaps and spectral techniques for embedding and clustering (2002)
Drawing on the correspondence between the graph Laplacian, the Laplace-Beltrami operator on a manifold, and the connections to the heat equation, we propose a geometrically motivated algorithm for...
Locality Preserving Projections (2002)
Many problems in information processing involve some form of dimensionality reduction. In this paper, we introduce Locality Preserving Projections (LPP). These are linear projective maps that arise...
Locality Preserving Projections (LPP (2002)
Many problems in information processing involve some form of dimensionality reduction. In this paper, we introduce Locality Preserving Projections (LPP). These are linear projective maps that arise...
Almost-everywhere algorithmic stability and generalization error (2002)
We introduce a new notion of algorithmic stability, which we call training stability. We show that training stability is sufficient for good bounds on generalization error. These bounds hold even...
Laplacian Eigenmaps for Dimensionality Reduction and Data Representation (2002)
One of the central problems in machine learning and pattern recognition is to develop appropriate representations for complex data. We consider the problem of constructing a representation for data...
Laplacian eigenmaps and spectral techniques for embedding and clustering (2002)
Drawing on the correspondence between the graph Laplacian, the Laplace-Beltrami operator on a manifold, and the connections to the heat equation, we propose a geometrically motivated algorithm for...
The evolutionary dynamics of grammar acquisition (2001)
Natalia L. Komarova, Partha Niyogi, A. Nowak
Grammar is the computational system of language. It is a set of rules that speci"es how to construct sentences out of words. Grammar is the basis of the unlimited expressibility of human...
Distinctive Feature Detection Using Support Vector Machines (1999)
Partha Niyogi, Chris Burges, Padma Ramesh
An important aspect of distinctive feature based approaches to automatic speech recognition is the formulation of a framework for robust detection of these features. We discuss the application of the...
Generalization bounds for function approximation from Scattered Noisy Data (1999)
Partha Niyogi, Federico Girosi
this paper we investigate the problem of providing error bounds for approximation of an unknown function from scattered, noisy data. This problem has particular relevance in the field of machine...
Modelling Speaker Variability and Imposing Speaker Constraints in Phonetic Classification. (1998)
This thesis deals with intraspeaker correlation analyses of speech sounds, and the possible utilization of this correlation to speech recognition. Current approaches to phonetic classification,...
Formalizing Triggers: A Learning Model for Finite Spaces. (1998)
Niyogi, Partha, Berwick, Robert C.
In a recent seminal paper, Gibson and Wexler (1993) take important steps to formalizing the notion of language teaming in a (finite) space whose grammars are characterized by a finite number of...
Sequential Optimal Recovery: A Paradigm for Active Learning. (1998)
In most classical frameworks for learning from examples, it is assumed that examples are randomly drawn and presented to the learner. In this paper, we consider the possibility of a more (it active)...
A Note on Zipf's Law, Natural Languages, and Noncoding DNA Regions. (1998)
Niyogi, Partha, Berwick, Robert C.
In Phys. Rev. Letiers, 73:2. 5 Dec. 94, Mantegna et al. conclude on the basis of Zipf rank frequency data that noncoding DNA sequence regions are more like natural languages than coding regions. We...
The Logical Problem of Language Change. (1998)
Niyogi, Partha, Berwick, Robert
This paper considers the problem of language change. Linguists must explain not only how languages are learned but also how and why they have evolved along certain trajectories and not others. While...
Generalization Bounds for Function Approximation from Scattered Noisy Data (1998)
Partha Niyogi, Federico Girosi
this paper we investigate the problem of providing error bounds for approximation of an unknown function from scattered, noisy data. This problem has particular relevance in the field of machine...
A Dynamical Systems Model for Language Change. (1997)
Niyogi, Partha, Berwick, Robert C.
Formalizing linguists' intuitions of language change as a dynamical system, we quantify the time course of language change including sudden vs. gradual changes in languages. We apply the computer...
Comparing Support Vector Machines with Gaussian Kernels to Radial Basis Function Classifiers (1997)
Bernhard Schölkopf, Kah-Kay Sung, Chris Burges, Federico Girosi, Partha Niyogi, Tomaso Poggio, ...
The Support Vector (SV) machine is a novel type of learning machine, based on statistical learning theory, which contains polynomial classifiers, neural networks, and radial basis function (RBF)...
The Informational Complexity of Learning from Examples (1996)
This thesis attempts to quantify the amount of information needed to learn certain tasks. The tasks chosen vary from learning functions in a Sobolev space using radial basis function networks to...
The Informational Complexity of Learning from Examples (1996)
This thesis attempts to quantify the amount of information needed to learn certain tasks. The tasks chosen vary from learning functions in a Sobolev space using radial basis function networks to...
A Formulation for Active Learning with Applications to Object Detection (1996)
We discuss a formulation for active example selection for function learning problems. This formulation is obtained by adapting Fedorov's optimal experiment design to the learning problem. We...
A Formulation for Active Learning with Applications to Object Detection (1996)
We discuss a formulation for active example selection for function learning problems. This formulation is obtained by adapting Fedorov's optimal experiment design to the learning problem. We...
Partha Niyogi, Federico Girosi
Feedforward networks are a class of regression techniques that can be used to learn to perform some task from a set of examples. The question of generalization of network performance from a finite...
The Informational Complexity of Learning from Examples (1996)
This thesis attempts to quantify the amount of information needed to learn certain tasks. The tasks chosen vary from learning functions in a Sobolev space using radial basis function networks to...
A Formulation for Active Learning with Applications to Object Detection (1996)
We discuss a formulation for active example selection for function learning problems. This formulation is obtained by adapting Fedorov's optimal experiment design to the learning problem. We...
Partha Niyogi, Federico Girosi
Feedforward networks together with their training algorithms are a class of regression techniques that can be used to learn to perform some task from a set of examples. The question of generalization...
The Logical Problem of Language Change (1995)
Niyogi, Partha, Berwick, Robert
This paper considers the problem of language change. Linguists must explain not only how languages are learned but also how and why they have evolved along certain trajectories and not others. While...
A Dynamical Systems Model for Language Change (1995)
Niyogi, Partha, Berwick, Robert
Formalizing linguists' intuitions of language change as a dynamical system, we quantify the time course of language change including sudden vs. gradual changes in languages. We apply the computer...
A Dynamical Systems Model for Language Change (1995)
Niyogi, Partha, Berwick, Robert
Formalizing linguists' intuitions of language change as a dynamical system, we quantify the time course of language change including sudden vs. gradual changes in languages. We apply the computer...
The Logical Problem of Language Change (1995)
Niyogi, Partha, Berwick, Robert
This paper considers the problem of language change. Linguists must explain not only how languages are learned but also how and why they have evolved along certain trajectories and not others. While...
Sequential Optimal Recovery: A Paradigm for Active Learning (1995)
In most classical frameworks for learning from examples, it is assumed that examples are randomly drawn and presented to the learner. In this paper, we consider the possibility of a more active...
Sequential Optimal Recovery: A Paradigm for Active Learning (1995)
In most classical frameworks for learning from examples, it is assumed that examples are randomly drawn and presented to the learner. In this paper, we consider the possibility of a more active...
A Note on Zipf's Law, Natural Languages, and Noncoding DNA regions (1995)
Niyogi, Partha, Berwick, Robert C.
In Phys. Rev. Letters (73:2, 5 Dec. 94), Mantegna et al. conclude on the basis of Zipf rank frequency data that noncoding DNA sequence regions are more like natural languages than coding regions. We...
A Note of Zipf's Law, Natural Languages, and Noncoding DNA Regions (1995)
Niyogi, Partha, Berwick, Robert C.
In Phys. Rev. Letters (73:2), Mantegna et al. conclude on the basis of Zipf rank frequency data that noncoding DNA sequence regions are more like natural languages than coding regions. We argue on...
A Note of Zipf's Law, Natural Languages, and Noncoding DNA Regions (1995)
Niyogi, Partha, Berwick, Robert C.
In Phys. Rev. Letters (73:2), Mantegna et al. conclude on the basis of Zipf rank frequency data that noncoding DNA sequence regions are more like natural languages than coding regions. We argue on...
The informational complexity of learning from examples /--by Partha Niyogi. (1995)
Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 1995.
The informational complexity of learning from examples (1995)
Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 1995.
The informational complexity of learning from examples (1995)
Thesis (Ph. D.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 1995.
The Logical Problem of Language Change (1995)
Partha Niyogi, Robert C. Berwick
This publication can be retrieved by anonymous ftp to publications.ai.mit.edu. This paper considers the problem of language change. Linguists must explain not only how languages are learned but also...
This publication can be retrieved by anonymous ftp to publications.ai.mit.edu. In Phys. Rev. Letters, 73:2, 5 Dec. 94, Mantegna et al. conclude on the basis of Zipf rank frequency data that noncoding...
Active Learning for Function Approximation (1995)
We develop a principled strategy to sample a function optimally for function approximation tasks within a Bayesian framework. Using ideas from optimal experiment design, we introduce an objective...
Active Learning The Weights Of A Rbf Network (1995)
We describe a principled strategy to sample functions optimally for function approximation tasks. The strategy works within a Bayesian framework and uses ideas from optimal experiment design to...
Active Learning by Sequential Optimal Recovery (1995)
In most classical frameworks for learning from examples, it is assumed that examples are randomly drawn and presented to the learner. In this paper, we consider the possibility of a more active...
Niyogi, Partha, Girosi, Federico
In this paper, we bound the generalization error of a class of Radial Basis Function networks, for certain well defined function learning tasks, in terms of the number of parameters and number of...
Niyogi, Partha, Girosi, Federico
In this paper, we bound the generalization error of a class of Radial Basis Function networks, for certain well defined function learning tasks, in terms of the number of parameters and number of...
Partha Niyogi, Federico Girosi
Feedforward networks are a class of approximation techniques that can be used to learn to perform some tasks from a finite set of examples. The question of the capability of a network to generalize...
Formalizing Triggers: A Learning Model for Finite Spaces (1993)
Niyogi, Partha, Berwick, Robert C.
In a recent seminal paper, Gibson and Wexler (1993) take important steps to formalizing the notion of language learning in a (finite) space whose grammars are characterized by a finite number of...
Formalizing Triggers: A Learning Model for Finite Spaces (1993)
Niyogi, Partha, Berwick, Robert C.
In a recent seminal paper, Gibson and Wexler (1993) take important steps to formalizing the notion of language learning in a (finite) space whose grammars are characterized by a finite number of...
Formalizing triggers: A learning model for finite spaces (1993)
Partha Niyogi, Robert C. Berwick
In a recent seminal paper, Gibson and Wexler ([1], GW) take important steps to formalizing the notion of language learning in a (finite) space whose grammars are characterized by a finite number of...
Cover title.
Modelling speaker variability and imposing speaker constraints in phonetic classification / (1991)
Thesis (M.S.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 1992.
A Dynamical Systems Model for Language Change
Partha Niyogi, Robert C. Berwick
Formalizing linguists' intuitions of language change as a dynamical system, we quantify the time course of language change including sudden vs. gradual changes in languages. We apply the...
The proper treatment of language acquisition and change in a population setting
Niyogi, Partha, Berwick, Robert C.
Language acquisition maps linguistic experience, primary linguistic data (PLD), onto linguistic knowledge, a grammar. Classically, computational models of language acquisition assume a single target...