Recent Google Publications (Atom)

Sampling Techniques for the Nystrom Method

2009-06-24T11:46:24-08:00

Sampling Techniques for the Nystrom Method, Sanjiv Kumar, Mehryar Mohri, Ameet Talwalkar

On Sampling-Based Approximate Spectral Decomposition

2009-06-24T11:47:16-08:00

On Sampling-Based Approximate Spectral Decomposition, Sanjiv Kumar, Mehryar Mohri, Ameet Talkwalkar

How opinions are received by online communities: A case study on Amazon.com helpfulness votes

2009-06-24T12:02:15-08:00

There are many on-line settings in which users publicly express opinions. A number of these offer mechanisms for other users to evaluate these opinions; a canonical example is Amazon.com, where reviews come with annotations like ``26 of 32 people found the following review helpful.'' Opinion evaluation appears in many off-line settings as well, including market research and political campaigns. Reasoning about the evaluation of an opinion is fundamentally different from reasoning about the opinion itself: rather than asking, ``What did Y think of X?'', we are asking, ``What did Z think of Y's opinion of X?'' Here we develop a framework for analyzing and modeling opinion evaluation, using a large-scale collection of Amazon book reviews as a dataset. We find that the perceived helpfulness of a review depends not just on its content but also but also in subtle ways on how the expressed evaluation relates to other evaluations of the same product. As part of our approach, we develop novel methods that take advantage of the phenomenon of review ``plagiarism'' to control for the effects of text in opinion evaluation, and we provide a simple and natural mathematical model consistent with our findings. Our analysis also allows us to distinguish among the predictions of competing theories from sociology and social psychology, and to discover unexpected differences in the collective opinion-evaluation behavior of user populations from different countries.

Web-Scale N-gram Models for Lexical Disambiguation

2009-06-24T12:03:13-08:00

Web-Scale N-gram Models for Lexical Disambiguation, Shane Bergsma, Dekang Lin, Randy Goebel

Glen, Glenda or Glendale: Unsupervised and Semi-supervised Learning of English Noun Gender

2009-06-24T11:32:48-08:00

Glen, Glenda or Glendale: Unsupervised and Semi-supervised Learning of English Noun Gender, Shane Bergsma, Dekang Lin, Randy Goebel

Online Learning with Global Cost Functions

2009-06-24T12:04:28-08:00

We consider an online learning setting where at each time step the decision maker has to choose how to distribute the future loss between k alternatives, and then observes the loss of each alternative. Motivated by load balancing and job scheduling, we consider a global cost function (over the losses incurred by each alternative), rather than a summation of the instantaneous losses as done traditionally in online learning. Such global cost functions include the makespan (the maximum over the alternatives) and the L_d norm (over the alternatives). Based on approachability theory, we design an algorithm that guarantees vanishing regret for this setting, where the regret is measured with respect to the best static decision that selects the same distribution over alternatives at every time step. For the special case of makespan cost we devise a simple and efficient algorithm. In contrast, we show that for concave global cost functions, such as L_d norms for d<1, the worst-case average regret does not vanish.

On the Convergence of Regret Minimization Dynamics in Concave Games

2009-06-24T11:59:43-08:00

We consider standard regret minimization setting where at each time step the decision maker has to choose a distribution over k alternatives, and then observes the loss of each alternative. The setting is very similar to the classical online job scheduling setting with three major differences:

Information model: in the regret minimization setting losses are only observed after the actions (assigning the job to a machine) is performed and not observed before the action selection, as assumed in the classical online job scheduling setting,
The comparison class: in regret minimization the comparison class is the best static algorithm (i.e., distribution over alternatives) and not the optimal offline solution.
Performance measure: In regret minimization we measure the additive difference to the optimal solution in the comparison class, in contrast to the ratio used in online job scheduling setting.

Motivated by load balancing and job scheduling, we consider a global cost function (over the losses incur by each alternative/machine), rather than simply a summation of the instantaneous losses as done traditionally in regret minimization. Such global cost functions include the makespan (the maximum over the alternatives/machines) and the L_d norm (over the alternatives/machines). The major contribution of this work is to design a novel regret minimization algorithm based on calibration that guarantees a vanishing average regret, where the regret is measured with respect to the best static decision maker, who selects the same distribution over alternatives at every time step. Our results hold for a wide class of global cost functions. which include the makespan and the L_d norms, for d>1. In contrast, we show that for concave global cost functions, such as L_d norms for d<1, the worst-case average regret does not vanish. In addition to the general calibration based algorithm, we provide simple and efficient algorithms for special interesting cases.

Quantum Annealing for Variational Bayes Inference

2009-06-24T12:05:36-08:00

Quantum Annealing for Variational Bayes Inference, Issei Sato, Kenichi Kurihara, Shu Tanaka, Seiji Miyashita, Hiroshi Nakagawa

Quantum Annealing for Clustering

2009-06-24T12:05:54-08:00

Quantum Annealing for Clustering, Kenichi Kurihara, Shu Tanaka, Seiji Miyashita

Sound Ranking Using Auditory Sparse-Code Representations

2009-06-24T12:08:03-08:00

The task of ranking sounds from text queries is a good test application for machine-hearing techniques, and particularly for comparison and evaluation of alternative sound representations in a large-scale setting. We have adapted a machine-vision system, ``passive-aggressive model for image retrieval'' (PAMIR), which efficiently learns, using a ranking-based cost function, a linear mapping from a very large sparse feature space to a large query-term space. Using this system allows us to focus on comparison of different auditory front ends and different ways of extracting sparse features from high-dimensional auditory images. In addition to two main auditory-image models, we also include and compare a family of more conventional MFCC front ends. The experimental results show a significant advantage for the auditory models over vector-quantized MFCCs. The two auditory models tested use the adaptive pole-zero filter cascade (PZFC) auditory filterbank and sparse-code feature extraction from stabilized auditory images via multiple vector quantizers. The models differ in their implementation of the strobed temporal integration used to generate the stabilized image. Using ranking precision-at-top-k performance measures, the best results are about 70% top-1 precision and 35% average precision, using a test corpus of thousands of sound files and a query vocabulary of hundreds of words.