Publications

This paper presents a unified framework for object detection, segmentation, and classification using regions. Region features are appealing in this context because: (1) they encode shape and scale information of objects naturally; (2) they are only mildly affected by background clutter. Regions have not been popular as features due to their sensitivity to segmentation errors. In this paper, we start by producing a robust bag of overlaid regions for each image using Arbeldez et al., CVPR 2009. Each region is represented by a rich set of image cues (shape, color and texture). We then learn region weights using a max-margin framework. In detection and segmentation, we apply a generalized Hough voting scheme to generate hypotheses of object locations, scales and support, followed by a verification classifier and a constrained segmenter on each hypothesis. The proposed approach significantly outperforms the state of the art on the ETHZ shape database(87.1% average detection rate compared to Ferrari et al. 's 67.2%), and achieves competitive performance on the Caltech 101 database.

Natural killer (NK) cells produce interferon (IFN)-γ and thus have been suggested to promote type I immunity during bacterial infections. Yet, Listeria monocytogenes (Lm) and some other pathogens encode proteins that cause increased NK cell activation. Here, we show that stimulation of NK cell activation increases susceptibility during Lm infection despite and independent from robust NK cell production of IFNγ. The increased susceptibility correlated with IL-10 production by responding NK cells. NK cells produced IL-10 as their IFNγ production waned and the Lm virulence protein p60 promoted induction of IL-10 production by mouse and human NK cells. NK cells consequently exerted regulatory effects to suppress accumulation and activation of inflammatory myeloid cells. Our results reveal new dimensions of the role played by NK cells during Lm infection and demonstrate the ability of this bacterial pathogen to exploit the induction of regulatory NK cell activity to increase host susceptibility.

ABSTRACT A string of recent studies has debated the exact form and physical origin of an evolutionary trend between the peak luminosity of Type Ia supernovae (SNe Ia) and the properties of the galaxies that host them. We shed new light on the discussion by presenting an analysis of ∼200 low-redshift SNe Ia in which we measure the separation of Hubble residuals (HR; as probes of luminosity) between two host-galaxy morphological types. We show that this separation can test the predictions made by recently proposed models, using an independently and empirically determined distribution of each morphological type in host-property space. Our results are partially consistent with the new HR–age slope, but we find significant scatter in the predictions from different galaxy catalogues. The inconsistency in age illuminates an issue in the current debate that was not obvious in the long-discussed mass models: HR–host-property models are strongly dependent on the methods employed to determine galaxy properties. While our results demonstrate the difficulty in constructing a universal model for age as a proxy for host environment, our results indeed identify evolutionary trends between mass, age, morphology, and HR values, encouraging (or requiring, if such trends are to be accounted for in cosmological studies) further investigation.

We review the development of virtual tests for high-temperature ceramic matrix composites with textile reinforcement. Success hinges on understanding the relationship between the microstructure of continuous-fiber composites, including its stochastic variability, and the evolution of damage events leading to failure. The virtual tests combine advanced experiments and theories to address physical, mathematical, and engineering aspects of material definition and failure prediction. Key new experiments include surface image correlation methods and synchrotron-based, micrometer-resolution 3D imaging, both executed at temperatures exceeding 1,500°C. Computational methods include new probabilistic algorithms for generating stochastic virtual specimens, as well as a new augmented finite element method that deals efficiently with arbitrary systems of crack initiation, bifurcation, and coalescence in heterogeneous materials. Conceptual advances include the use of topology to characterize stochastic microstructures. We discuss the challenge of predicting the probability of an extreme failure event in a computationally tractable manner while retaining the necessary physical detail.

Abstract not Available.

We report the design of surfaces that exhibit dynamic changes in interfacial properties, such as wettability, in response to an electrical potential. The change in wetting behavior was caused by surface-confined, single-layered molecules undergoing conformational transitions between a hydrophilic and a moderately hydrophobic state. Reversible conformational transitions were confirmed at a molecular level with the use of sum-frequency generation spectroscopy and at a macroscopic level with the use of contact angle measurements. This type of surface design enables amplification of molecular-level conformational transitions to macroscopic changes in surface properties without altering the chemical identity of the surface. Such reversibly switching surfaces may open previously unknown opportunities in interfacial engineering.

This paper asks whether inter-flow contention between congestion control algorithms (CCAs) is a dominant factor in determining a flow's bandwidth allocation in today's Internet. We hypothesize that CCA contention typically does not determine a flow's bandwidth allocation, present an initial analysis in support of this hypothesis, propose a measurement technique and study to settle this question, and discuss the implications should the hypothesis prove true.

Formaldehyde (FA) is both a highly reactive environmental genotoxin and an endogenously produced metabolite that functions as a signaling molecule and one-carbon (1C) store to regulate 1C metabolism and epigenetics in the cell. Owing to its signal-stress duality, cells have evolved multiple clearance mechanisms to maintain FA homeostasis, acting to avoid the established genotoxicity of FA while also redirecting FA-derived carbon units into the biosynthesis of essential nucleobases and amino acids. The highly compartmentalized nature of FA exposure, production, and regulation motivates the development of chemical tools that enable monitoring of transient FA fluxes with subcellular resolution. Here we report a mitochondrial-targeted, activity-based sensing probe for ratiometric FA detection, MitoRFAP-2, and apply this reagent to monitor endogenous mitochondrial sources and sinks of this 1C unit. We establish the utility of subcellular localization by showing that MitoRFAP-2 is sensitive enough to detect changes in mitochondrial FA pools with genetic and pharmacological modulation of enzymes involved in 1C and amino acid metabolism, including the pervasive, less active genetic mutant aldehyde dehydrogenase 2*2 (ALDH2*2), where previous, non-targeted versions of FA sensors are not. Finally, we used MitoRFAP-2 to comparatively profile basal levels of FA across a panel of breast cancer cell lines, finding that FA-dependent fluorescence correlates with expression levels of enzymes involved in 1C metabolism. By showcasing the ability of MitoRFAP-2 to identify new information on mitochondrial FA homeostasis, this work provides a starting point for the design of a broader range of chemical probes for detecting physiologically important aldehydes with subcellular resolution and a useful reagent for further studies of 1C biology.