Showing papers by "Cornell University published in 2004"

Finding and evaluating community structure in networks.

Abstract: We propose and study a set of algorithms for discovering community structure in networks-natural divisions of network nodes into densely connected subgroups. Our algorithms all share two definitive features: first, they involve iterative removal of edges from the network to split it into communities, the edges removed being identified using any one of a number of possible "betweenness" measures, and second, these measures are, crucially, recalculated after each removal. We also propose a measure for the strength of the community structure found by our algorithms, which gives us an objective metric for choosing the number of communities into which a network should be divided. We demonstrate that our algorithms are highly effective at discovering community structure in both computer-generated and real-world network data, and show how they can be used to shed light on the sometimes dauntingly complex structure of networked systems.

Efficient Graph-Based Image Segmentation

Abstract: This paper addresses the problem of segmenting an image into regions. We define a predicate for measuring the evidence for a boundary between two regions using a graph-based representation of the image. We then develop an efficient segmentation algorithm based on this predicate, and show that although this algorithm makes greedy decisions it produces segmentations that satisfy global properties. We apply the algorithm to image segmentation using two different kinds of local neighborhoods in constructing the graph, and illustrate the results with both real and synthetic images. The algorithm runs in time nearly linear in the number of graph edges and is also fast in practice. An important characteristic of the method is its ability to preserve detail in low-variability image regions while ignoring detail in high-variability regions.

Nitrogen cycles: past, present, and future

Abstract: This paper contrasts the natural and anthropogenic controls on the conversion of unreactive N2 to more reactive forms of nitrogen (Nr). A variety of data sets are used to construct global N budgets for 1860 and the early 1990s and to make projections for the global N budget in 2050. Regional N budgets for Asia, North America, and other major regions for the early 1990s, as well as the marine N budget, are presented to highlight the dominant fluxes of nitrogen in each region. Important findings are that human activities increasingly dominate the N budget at the global and at most regional scales, the terrestrial and open ocean N budgets are essentially dis- connected, and the fixed forms of N are accumulating in most environmental reservoirs. The largest uncertainties in our understanding of the N budget at most scales are the rates of natural biological nitrogen fixation, the amount of Nr storage in most environmental reservoirs, and the production rates of N2 by denitrification.

The Gene Ontology (GO) database and informatics resource.

Abstract: The Gene Ontology (GO) project (http://www.geneontology.org/) provides structured, controlled vocabularies and classifications that cover several domains of molecular and cellular biology and are freely available for community use in the annotation of genes, gene products and sequences. Many model organism databases and genome annotation groups use the GO and contribute their annotation sets to the GO resource. The GO database integrates the vocabularies and contributed annotations and provides full access to this information in several formats. Members of the GO Consortium continually work collectively, involving outside experts as needed, to expand and update the GO vocabularies. The GO Web resource also provides access to extensive documentation about the GO project and links to applications that use GO data for functional analyses.

A Sentimental Education: Sentiment Analysis Using Subjectivity Summarization Based on Minimum Cuts

Abstract: Sentiment analysis seeks to identify the viewpoint(s) underlying a text span; an example application is classifying a movie review as "thumbs up" or "thumbs down". To determine this sentiment polarity, we propose a novel machine-learning method that applies text-categorization techniques to just the subjective portions of the document. Extracting these portions can be implemented using efficient techniques for finding minimum cuts in graphs; this greatly facilitates incorporation of cross-sentence contextual constraints.

Trichoderma species--opportunistic, avirulent plant symbionts.

Abstract: Trichoderma spp. are free-living fungi that are common in soil and root ecosystems. Recent discoveries show that they are opportunistic, avirulent plant symbionts, as well as being parasites of other fungi. At least some strains establish robust and long-lasting colonizations of root surfaces and penetrate into the epidermis and a few cells below this level. They produce or release a variety of compounds that induce localized or systemic resistance responses, and this explains their lack of pathogenicity to plants. These root-microorganism associations cause substantial changes to the plant proteome and metabolism. Plants are protected from numerous classes of plant pathogen by responses that are similar to systemic acquired resistance and rhizobacteria-induced systemic resistance. Root colonization by Trichoderma spp. also frequently enhances root growth and development, crop productivity, resistance to abiotic stresses and the uptake and use of nutrients.

The Spitzer Space Telescope mission

Abstract: The Spitzer Space Telescope, NASA's Great Observatory for infrared astronomy, was launched 2003 August 25 and is returning excellent scientific data from its Earth-trailing solar orbit. Spitzer combines the intrinsic sensitivity achievable with a cryogenic telescope in space with the great imaging and spectroscopic power of modern detector arrays to provide the user community with huge gains in capability for exploration of the cosmos in the infrared. The observatory systems are largely performing as expected, and the projected cryogenic lifetime is in excess of 5 years. This paper summarizes the on-orbit scientific, technical, and operational performance of Spitzer. Subsequent papers in this special issue describe the Spitzer instruments in detail and highlight many of the exciting scientific results obtained during the first 6 months of the Spitzer mission.

What energy functions can be minimized via graph cuts

Abstract: In the last few years, several new algorithms based on graph cuts have been developed to solve energy minimization problems in computer vision. Each of these techniques constructs a graph such that the minimum cut on the graph also minimizes the energy. Yet, because these graph constructions are complex and highly specific to a particular energy function, graph cuts have seen limited application to date. In this paper, we give a characterization of the energy functions that can be minimized by graph cuts. Our results are restricted to functions of binary variables. However, our work generalizes many previous constructions and is easily applicable to vision problems that involve large numbers of labels, such as stereo, motion, image restoration, and scene reconstruction. We give a precise characterization of what energy functions can be minimized using graph cuts, among the energy functions that can be written as a sum of terms containing three or fewer binary variables. We also provide a general-purpose construction to minimize such an energy function. Finally, we give a necessary condition for any energy function of binary variables to be minimized by graph cuts. Researchers who are considering the use of graph cuts to optimize a particular energy function can use our results to determine if this is possible and then follow our construction to create the appropriate graph. A software implementation is freely available.

The Cross-Section of Volatility and Expected Returns

Abstract: We examine the pricing of aggregate volatility risk in the cross-section of stock returns. Consistent with theory, we find that stocks with high sensitivities to innovations in aggregate volatility have low average returns. In addition, we find that stocks with high idiosyncratic volatility relative to the Fama and French (1993) model have abysmally low average returns. This phenomenon cannot be explained by exposure to aggregate volatility risk. Size, book-to-market, momentum, and liquidity effects cannot account for either the low average returns earned by stocks with high exposure to systematic volatility risk or for the low average returns of stocks with high idiosyncratic volatility.

Potential Synergy of Phytochemicals in Cancer Prevention: Mechanism of Action

Abstract: Epidemiological studies have consistently shown that regular consumption of fruits and vegetables is strongly associated with reduced risk of developing chronic diseases, such as cancer and cardiovascular disease. It is now widely believed that the actions of the antioxidant nutrients alone do not explain the observed health benefits of diets rich in fruits and vegetables, because taken alone, the individual antioxidants studied in clinical trials do not appear to have consistent preventive effects. Work performed by our group and others has shown that fruits and vegetable phytochemical extracts exhibit strong antioxidant and antiproliferative activities and that the major part of total antioxidant activity is from the combination of phytochemicals. We proposed that the additive and synergistic effects of phytochemicals in fruits and vegetables are responsible for these potent antioxidant and anticancer activities and that the benefit of a diet rich in fruits and vegetables is attributed to the complex mixture of phytochemicals present in whole foods. This explains why no single antioxidant can replace the combination of natural phytochemicals in fruits and vegetables to achieve the health benefits. The evidence suggests that antioxidants or bioactive compounds are best acquired through whole-food consumption, not from expensive dietary supplements. We believe that a recommendation that consumers eat 5 to 10 servings of a wide variety of fruits and vegetables daily is an appropriate strategy for significantly reducing the risk of chronic diseases and to meet their nutrient requirements for optimum health.

Neurovascular regulation in the normal brain and in Alzheimer's disease.

Abstract: The structural and functional integrity of the brain depends on the delicate balance between substrate delivery through blood flow and energy demands imposed by neural activity. Complex cerebrovascular control mechanisms ensure that active brain regions receive an adequate amount of blood, but the nature of these mechanisms remains elusive. Recent findings implicate perivascular neurons, gliovascular interactions and intramural vascular signalling in the control of the cerebral microcirculation. Neurons, astrocytes and vascular cells seem to constitute a functional unit, the primary purpose of which is to maintain the homeostasis of the brain's microenvironment. Alterations of these vascular regulatory mechanisms lead to brain dysfunction and disease. The emerging view is that cerebrovascular dysregulation is a feature not only of cerebrovascular pathologies, such as stroke, but also of neurodegenerative conditions, such as Alzheimer's disease.

Global Mapping of the Yeast Genetic Interaction Network

Abstract: A genetic interaction network containing approximately 1000 genes and approximately 4000 interactions was mapped by crossing mutations in 132 different query genes into a set of approximately 4700 viable gene yeast deletion mutants and scoring the double mutant progeny for fitness defects. Network connectivity was predictive of function because interactions often occurred among functionally related genes, and similar patterns of interactions tended to identify components of the same pathway. The genetic network exhibited dense local neighborhoods; therefore, the position of a gene on a partially mapped network is predictive of other genetic interactions. Because digenic interactions are common in yeast, similar networks may underlie the complex genetics associated with inherited phenotypes in other organisms.

The Classification of Glomerulonephritis in Systemic Lupus Erythematosus Revisited

Abstract: The currently used classification reflects our understanding of the pathogenesis of the various forms of lupus nephritis, but clinicopathologic studies have revealed the need for improved categorization and terminology. Based on the 1982 classification published under the auspices of the World Health Organization (WHO) and subsequent clinicopathologic data, we propose that class I and II be used for purely mesangial involvement (I, mesangial immune deposits without mesangial hypercellularity; II, mesangial immune deposits with mesangial hypercellularity); class III for focal glomerulonephritis (involving or = 50% of total number of glomeruli) either with segmental (class IV-S) or global (class IV-G) involvement, and also with subdivisions for active and sclerotic lesions; class V for membranous lupus nephritis; and class VI for advanced sclerosing lesions]. Combinations of membranous and proliferative glomerulonephritis (i.e., class III and V or class IV and V) should be reported individually in the diagnostic line. The diagnosis should also include entries for any concomitant vascular or tubulointerstitial lesions. One of the main advantages of the current revised classification is that it provides a clear and unequivocal description of the various lesions and classes of lupus nephritis, allowing a better standardization and lending a basis for further clinicopathologic studies. We hope that this revision, which evolved under the auspices of the International Society of Nephrology and the Renal Pathology Society, will contribute to further advancement of the WHO classification.

Enhancing Treatment Fidelity in Health Behavior Change Studies: Best Practices and Recommendations From the NIH Behavior Change Consortium.

Abstract: Treatment fidelity refers to the methodological strategies used to monitor and enhance the reliability and validity of behavioral interventions. This article describes a multisite effort by the Treatment Fidelity Workgroup of the National Institutes of Health Behavior Change Consortium (BCC) to identify treatment fidelity concepts and strategies in health behavior intervention research. The work group reviewed treatment fidelity practices in the research literature, identified techniques used within the BCC, and developed recommendations for incorporating these practices more consistently. The recommendations cover study design, provider training, treatment delivery, treatment receipt, and enactment of treatment skills. Funding agencies, reviewers, and journal editors are encouraged to make treatment fidelity a standard part of the conduct and evaluation of health behavior intervention research.

A Social Influence Model of Consumer Participation in Network- and Small-Group-Based Virtual Communities

Abstract: We investigate two key group-level determinants of virtual community participation - group norms and social identity - and consider their motivational antecedents and mediators. We also introduce a marketing-relevant typology to conceptualize virtual communities, based on the distinction between network-based, and small-group-based virtual communities. Our survey-based study, which was conducted across a broad range of virtual communities, supports the proposed model, and finds further that virtual community type moderates consumers' reasons for participating, as well as the strengths of their impact on group norms and social identity. We conclude with a consideration of managerial and research implications of the findings.

Vortex-induced vibrations

Abstract: This review summarizes fundamental results and discoveries concerning vortex-induced vibration (VIV), that have been made over the last two decades, many of which are related to the push to explore very low mass and damping, and to new computational and experimental techniques that were hitherto not available. We bring together new concepts and phenomena generic to VIV systems, and pay special attention to the vortex dynamics and energy transfer that give rise to modes of vibration, the importance of mass and damping, the concept of a critical mass, the relationship between force and vorticity, and the concept of "effective elasticity," among other points. We present new vortex wake modes, generally in the framework of a map of vortex modes compiled from forced vibration studies, some of which cause free vibration. Some discussion focuses on topics of current debate, such as the decomposition of force, the relevance of the paradigm flow of an elastically mounted cylinder to more complex systems, and the relationship between forced and free vibration.

The environment of childhood poverty.

Abstract: Poor children confront widespread environmental inequities. Compared with their economically advantaged counterparts, they are exposed to more family turmoil, violence, separation from their families, instability, and chaotic households. Poor children experience less social support, and their parents are less responsive and more authoritarian. Low-income children are read to relatively infrequently, watch more TV, and have less access to books and computers. Low-income parents are less involved in their children's school activities. The air and water poor children consume are more polluted. Their homes are more crowded, noisier, and of lower quality. Low-income neighborhoods are more dangerous, offer poorer municipal services, and suffer greater physical deterioration. Predominantly low-income schools and day care are inferior. The accumulation of multiple environmental risks rather than singular risk exposure may be an especially pathogenic aspect of childhood poverty.

Practice Guidelines for the Management of Bacterial Meningitis

Abstract: Allan R. Tunkel, Barry J. Hartman, Sheldon L. Kaplan, Bruce A. Kaufman, Karen L. Roos, W. Michael Scheld, and Richard J. Whitley Drexel University College of Medicine, Philadelphia, Pennsylvania; Weill Cornell Medical Center, New York, New York; Baylor College of Medicine, Houston, Texas; Medical College of Wisconsin, Milwaukee; Indiana University School of Medicine, Indianapolis; University of Virginia School of Medicine, Charlottesville; and University of Alabama at Birmingham

Sampling and Estimation in Hidden Populations Using Respondent-Driven Sampling

Abstract: Standard statistical methods often provide no way to make accurate estimates about the characteristics of hidden populations such as injection drug users, the homeless, and artists. In this paper, we further develop a sampling and estimation technique called respondent-driven sampling, which allows researchers to make asymptotically unbiased estimates about these hidden populations. The sample is selected with a snowball-type design that can be done more cheaply, quickly, and easily than other methods currently in use. Further, we can show that under certain specified (and quite general) conditions, our estimates for the percentage of the population with a specific trait are asymptotically unbiased. We further show that these estimates are asymptotically unbiased no matter how the seeds are selected. We conclude with a comparison of respondent-driven samples of jazz musicians in New York and San Francisco, with corresponding institutional samples of jazz musicians from these cities. The results show that ...

Guiding and confining light in void nanostructure.

Abstract: We present a novel waveguide geometry for enhancing and confining light in a nanometer-wide low-index material. Light enhancement and confinement is caused by large discontinuity of the electric field at highindex-contrast interfaces. We show that by use of such a structure the field can be confined in a 50-nm-wide low-index region with a normalized intensity of 20 mm 22 . This intensity is approximately 20 times higher than what can be achieved in SiO2 with conventional rectangular waveguides. © 2004 Optical Society of America OCIS codes: 030.4070, 130.0130, 130.2790, 230.7370, 230.7380, 230.7390, 230.7400. Recent results in integrated optics have shown the ability to guide, bend, split, and f ilter light on chips by use of optical devices based on high-index-contrast waveguides. 1–5 In all these devices the guiding mechanism is based on total internal ref lection (TIR) in a highindex material (core) surrounded by a low-indexmaterial (cladding); the TIR mechanism can strongly confine light in the high-index material. In recent years a number of structures have been proposed to guide or enhance light in low-index materials, 6–1 1 relying on external ref lections provided by interference effects. Unlike TIR, the external ref lection cannot be perfectly unity; therefore the modes in these structures are inherently leaky modes. In addition, since interference is involved, these structures are strongly wavelength dependent. Here we show that the optical field can be enhanced and conf ined in the low-index material even when light is guided by TIR. For a high-index-contrast interface, Maxwell’s equations state that, to satisfy the continuity of the normal component of electric f lux density D, the corresponding electric field (E-field) must undergo a large discontinuity with much higher amplitude in the low-index side. We show that this discontinuity can be used to strongly enhance and confine light in a nanometer-wide region of low-index material. The proposed structure presents an eigenmode, and it is compatible with highly integrated photonics technology. The principle of operation of the novel structure can be illustrated by analysis of the slab-based structure shown in Fig. 1(a), where a low-index slot is embedded between two high-index slabs (shaded regions). The novel structure is hereafter referred to as a slot waveguide. The slot waveguide eigenmode can be seen as being formed by the interaction between the fundamental eigenmodes of the individual slab waveguides. Rigorously, the analytical solution for the transverse E-field profile Ex of the fundamental TM eigenmode of the slab-based slot waveguide is

The infrared spectrograph (irs) on the spitzer space telescope

Abstract: The Infrared Spectrograph (IRS) is one of three science instruments on the Spitzer Space Telescope .T he IRS comprises four separate spectrograph modules covering the wavelength range from 5.3 to 38 � m with spectral resolutions, R ¼ k=� k � 90 and 600, and it was optimized to take full advantage of the very low background in the space environment. The IRS is performing at or better than the prelaunch predictions. An autonomous target acquisition capability enables the IRS to locate the mid-infrared centroid of a source, providing the information so that the spacecraft can accurately offset that centroid to a selected slit. This feature is particularly useful when taking spectra of sources with poorly known coordinates. An automated data-reduction pipeline has been developed at the Spitzer Science Center. Subject headingg infrared: general — instrumentation: spectrographs — space vehicles: instruments

Flawed Self-Assessment Implications for Health, Education, and the Workplace

Abstract: Research from numerous corners of psychological inquiry suggests that self-assessments of skill and character are often flawed in substantive and systematic ways. We review empirical findings on the imperfect nature of self-assessment and discuss implications for three real-world domains: health, education, and the workplace. In general, people's self-views hold only a tenuous to modest relationship with their actual behavior and performance. The correlation between self-ratings of skill and actual performance in many domains is moderate to meager-indeed, at times, other people's predictions of a person's outcomes prove more accurate than that person's self-predictions. In addition, people overrate themselves. On average, people say that they are "above average" in skill (a conclusion that defies statistical possibility), overestimate the likelihood that they will engage in desirable behaviors and achieve favorable outcomes, furnish overly optimistic estimates of when they will complete future projects, and reach judgments with too much confidence. Several psychological processes conspire to produce flawed self-assessments. Research focusing on health echoes these findings. People are unrealistically optimistic about their own health risks compared with those of other people. They also overestimate how distinctive their opinions and preferences (e.g., discomfort with alcohol) are among their peers-a misperception that can have a deleterious impact on their health. Unable to anticipate how they would respond to emotion-laden situations, they mispredict the preferences of patients when asked to step in and make treatment decisions for them. Guided by mistaken but seemingly plausible theories of health and disease, people misdiagnose themselves-a phenomenon that can have severe consequences for their health and longevity. Similarly, research in education finds that students' assessments of their performance tend to agree only moderately with those of their teachers and mentors. Students seem largely unable to assess how well or poorly they have comprehended material they have just read. They also tend to be overconfident in newly learned skills, at times because the common educational practice of massed training appears to promote rapid acquisition of skill-as well as self-confidence-but not necessarily the retention of skill. Several interventions, however, can be introduced to prompt students to evaluate their skill and learning more accurately. In the workplace, flawed self-assessments arise all the way up the corporate ladder. Employees tend to overestimate their skill, making it difficult to give meaningful feedback. CEOs also display overconfidence in their judgments, particularly when stepping into new markets or novel projects-for example, proposing acquisitions that hurt, rather then help, the price of their company's stock. We discuss several interventions aimed at circumventing the consequences of such flawed assessments; these include training people to routinely make cognitive repairs correcting for biased self-assessments and requiring people to justify their decisions in front of their peers. The act of self-assessment is an intrinsically difficult task, and we enumerate several obstacles that prevent people from reaching truthful self-impressions. We also propose that researchers and practitioners should recognize self-assessment as a coherent and unified area of study spanning many subdisciplines of psychology and beyond. Finally, we suggest that policymakers and other people who makes real-world assessments should be wary of self-assessments of skill, expertise, and knowledge, and should consider ways of repairing self-assessments that may be flawed.

The Infrared Spectrograph on the Spitzer Space Telescope

Abstract: The Infrared Spectrograph (IRS) is one of three science instruments on the Spitzer Space Telescope. The IRS comprises four separate spectrograph modules covering the wavelength range from 5.3 to 38micron with spectral resolutions, R \~90 and 600, and it was optimized to take full advantage of the very low background in the space environment. The IRS is performing at or better than the pre-launch predictions. An autonomous target acquisition capability enables the IRS to locate the mid-infrared centroid of a source, providing the information so that the spacecraft can accurately offset that centroid to a selected slit. This feature is particularly useful when taking spectra of sources with poorly known coordinates. An automated data reduction pipeline has been developed at the Spitzer Science Center.

How do crop plants tolerate acid soils? Mechanisms of aluminum tolerance and phosphorous efficiency.

Abstract: Acid soils significantly limit crop production worldwide because approximately 50% of the world's potentially arable soils are acidic. Because acid soils are such an important constraint to agriculture, understanding the mechanisms and genes conferring tolerance to acid soil stress has been a focus of intense research interest over the past decade. The primary limitations on acid soils are toxic levels of aluminum (Al) and manganese (Mn), as well as suboptimal levels of phosphorous (P). This review examines our current understanding of the physiological, genetic, and molecular basis for crop Al tolerance, as well as reviews the emerging area of P efficiency, which involves the genetically based ability of some crop genotypes to tolerate P deficiency stress on acid soils. These are interesting times for this field because researchers are on the verge of identifying some of the genes that confer Al tolerance in crop plants; these discoveries will open up new avenues of molecular/physiological inquiry that should greatly advance our understanding of these tolerance mechanisms. Additionally, these breakthroughs will provide new molecular resources for improving crop Al tolerance via both molecular-assisted breeding and biotechnology.

All-optical control of light on a silicon chip

Abstract: Photonic circuits, in which beams of light redirect the flow of other beams of light, are a long-standing goal for developing highly integrated optical communication components1,2,3. Furthermore, it is highly desirable to use silicon—the dominant material in the microelectronic industry—as the platform for such circuits. Photonic structures that bend, split, couple and filter light have recently been demonstrated in silicon4,5, but the flow of light in these structures is predetermined and cannot be readily modulated during operation. All-optical switches and modulators have been demonstrated with III–V compound semiconductors6,7, but achieving the same in silicon is challenging owing to its relatively weak nonlinear optical properties. Indeed, all-optical switching in silicon has only been achieved by using extremely high powers8,9,10,11,12,13,14,15 in large or non-planar structures, where the modulated light is propagating out-of-plane. Such high powers, large dimensions and non-planar geometries are inappropriate for effective on-chip integration. Here we present the experimental demonstration of fast all-optical switching on silicon using highly light-confining structures to enhance the sensitivity of light to small changes in refractive index. The transmission of the structure can be modulated by up to 94% in less than 500 ps using light pulses with energies as low as 25 pJ. These results confirm the recent theoretical prediction16 of efficient optical switching in silicon using resonant structures.

Multilocus Methods for Estimating Population Sizes, Migration Rates and Divergence Time, With Applications to the Divergence of Drosophila pseudoobscura and D. persimilis

Abstract: The genetic study of diverging, closely related populations is required for basic questions on demography and speciation, as well as for biodiversity and conservation research. However, it is often unclear whether divergence is due simply to separation or whether populations have also experienced gene flow. These questions can be addressed with a full model of population separation with gene flow, by applying a Markov chain Monte Carlo method for estimating the posterior probability distribution of model parameters. We have generalized this method and made it applicable to data from multiple unlinked loci. These loci can vary in their modes of inheritance, and inheritance scalars can be implemented either as constants or as parameters to be estimated. By treating inheritance scalars as parameters it is also possible to address variation among loci in the impact via linkage of recurrent selective sweeps or background selection. These methods are applied to a large multilocus data set from Drosophila pseudoobscura and D. persimilis. The species are estimated to have diverged ∼500,000 years ago. Several loci have nonzero estimates of gene flow since the initial separation of the species, with considerable variation in gene flow estimates among loci, in both directions between the species.

Support vector machine learning for interdependent and structured output spaces

Abstract: Learning general functional dependencies is one of the main goals in machine learning. Recent progress in kernel-based methods has focused on designing flexible and powerful input representations. This paper addresses the complementary issue of problems involving complex outputs such as multiple dependent output variables and structured output spaces. We propose to generalize multiclass Support Vector Machine learning in a formulation that involves features extracted jointly from inputs and outputs. The resulting optimization problem is solved efficiently by a cutting plane algorithm that exploits the sparseness and structural decomposition of the problem. We demonstrate the versatility and effectiveness of our method on problems ranging from supervised grammar learning and named-entity recognition, to taxonomic text classification and sequence alignment.