Global Contextual Dependency Network for Object Detection

Current two-stage object detectors extract the local visual features of Regions of Interest (RoIs) for object recognition and bounding-box regression. However, only using local visual features will lose global contextual dependencies, which are helpful to recognize objects with featureless appearances and restrain false detections. To tackle the problem, a simple framework, named Global Contextual Dependency Network (GCDN), is presented to enhance the classification ability of two-stage detectors. Our GCDN mainly consists of two components, Context Representation Module (CRM) and Context Dependency Module (CDM). Specifically, a CRM is proposed to construct multi-scale context representations. With CRM, contextual information can be fully explored at different scales. Moreover, the CDM is designed to capture global contextual dependencies. Our GCDN includes multiple CDMs. Each CDM utilizes local Region of Interest (RoI) features and single-scale context representation to generate single-scale contextual RoI features via the attention mechanism. Finally, the contextual RoI features generated by parallel CDMs independently are combined with the original RoI features to help classification. Experiments on MS-COCO 2017 benchmark dataset show that our approach brings continuous improvements for two-stage detectors.

Epichloë Endophyte-Promoted Seed Pathogen Increases Host Grass Resistance Against Insect Herbivory

Plants host taxonomically and functionally complex communities of microbes. However, ecological studies on plant–microbe interactions rarely address the role of multiple co-occurring plant-associated microbes. Here, we contend that plant-associated microbes interact with each other and can have joint consequences for higher trophic levels. In this study we recorded the occurrence of the plant seed pathogenic fungus Claviceps purpurea and aphids (Sitobion sp.) on an established field experiment with red fescue (Festuca rubra) plants symbiotic to a seed transmitted endophytic fungus Epichloë festucae (E+) or non-symbiotic (E–). Both fungi are known to produce animal-toxic alkaloids. The study was conducted in a semi-natural setting, where E+ and E– plants from different origins (Spain and Northern Finland) were planted in a randomized design in a fenced common garden at Kevo Subarctic Research Station in Northern Finland. The results reveal that 45% of E+ plants were infected with Claviceps compared to 31% of E– plants. Uninfected plants had 4.5 times more aphids than Claviceps infected plants. By contrast, aphid infestation was unaffected by Epichloë symbiosis. Claviceps alkaloid concentrations correlated with a decrease in aphid numbers, which indicates their insect deterring features. These results show that plant mutualistic fungi can increase the infection probability of a pathogenic fungus, which then becomes beneficial to the plant by controlling herbivorous insects. Our study highlights the complexity and context dependency of species–species and multi-trophic interactions, thus challenging the labeling of species as plant mutualists or pathogens.

Defense in Social Insects: Diversity, Division of Labor, and Evolution

All social insects defend their colony from predators, parasites, and pathogens. In Oster and Wilson's classic work, they posed one of the key paradoxes about defense in social insects: Given the universal necessity of defense, why then is there so much diversity in mechanisms? Ecological factors undoubtedly are important: Predation and usurpation have imposed strong selection on eusocial insects, and active defense by colonies is a ubiquitous feature of all social insects. The description of diverse insect groups with castes of sterile workers whose main duty is defense has broadened the purview of social evolution in insects, in particular with respect to caste and behavior. Defense is one of the central axes along which we can begin to organize and understand sociality in insects. With the establishment of social insect models such as the honey bee, new discoveries are emerging regarding the endocrine, neural, and gene regulatory mechanisms underlying defense in social insects. The mechanisms underlying morphological and behavioral defense traits may be shared across diverse groups, providing opportunities for identifying both conserved and novel mechanisms at work. Emerging themes highlight the context dependency of and interaction between factors that regulate defense in social insects.

Maintenance of Intraspecific Diversity in Response to Species Competition and Nutrient Fluctuations

Intraspecific diversity is a substantial part of biodiversity, yet little is known about its maintenance. Understanding mechanisms of intraspecific diversity shifts provides realistic detail about how phytoplankton communities evolve to new environmental conditions, a process especially important in times of climate change. Here, we aimed to identify factors that maintain genotype diversity and link the observed diversity change to measured phytoplankton morpho-functional traits Vmax and cell size of the species and genotypes. In an experimental setup, the two phytoplankton species Emiliania huxleyi and Chaetoceros affinis, each consisting of nine genotypes, were cultivated separately and together under different fluctuation and nutrient regimes. Their genotype composition was assessed after 49 and 91 days, and Shannon’s diversity index was calculated on the genotype level. We found that a higher intraspecific diversity can be maintained in the presence of a competitor, provided it has a substantial proportion to total biovolume. Both fluctuation and nutrient regime showed species-specific effects and especially structured genotype sorting of C. affinis. While we could relate species sorting with the measured traits, genotype diversity shifts could only be partly explained. The observed context dependency of genotype maintenance suggests that the evolutionary potential could be better understood, if studied in more natural settings including fluctuations and competition.

Contextual restrictions on cumulativity

This paper revisits the semantic variability of sentences with simple plural (in)definites in English and German, which permit distributive, cumulative and paired-cover construals. I argue that this variability reflects context-dependency rather than LF ambiguity (Schwarzschild 1996) and that the selection of a particular construal in context is driven by the QUD in the same way as the choice between maximal and non-maximal construals of plural definites (Malamud 2012; Križ 2015; Križ & Spector 2020). I then develop a new semantics for plural predication on which non-distributive and non-maximal construals form a natural class. The system extends the idea that non-maximality involves truth-value gaps (e.g. Križ 2015) to non-distributive construals by making use of Schmitt’s (2019) ‘plural projection’ framework, in which plural sentences involve special composition rules.

Mutant KRAS-associated proteome is mainly controlled by exogenous factors

KRAS signaling has been extensively studied, yet the clarification between KRAS-autonomous and non-autonomous mechanisms are still less explored. Understanding how KRAS signaling and effects are affected by exogenous stimuli can provide valuable insights not only to understand resistance mechanisms that justify pathway inhibition failure, but also to uncover novel therapeutic targets for mutant KRAS patients. Hence, aiming at understanding KRAS-autonomous versus non autonomous mechanisms, we studied the response of two mutant KRAS colorectal cancer cell lines (HCT116 and LS174T) - control and KRAS silenced- to TGFβ1-activated fibroblasts secretome. By performing a total proteome analysis, we observed that TGFβ1-activated fibroblast-secreted factors triggered cell line-specific proteome alterations and that mutant KRAS governs approximately 1/3 of those alterations. Moreover, the analysis of the impact of exogenous factors on the modulation of KRAS proteome revealed that, in both cell lines, more than 2/3 of the KRAS-associated proteome is controlled in a KRAS-non-autonomous manner and dependent on the exogenous factors. This work highlights the context-dependency of KRAS-associated signaling and reinforces the importance of establishing more integrative models resembling the complexity of the tumor microenvironment to study KRAS-associated signals.

A multivariate modeling framework to quantify immune checkpoint context-dependent stimulation on T cells

Cells receive, and adjust to, various stimuli, which function as part of complex microenvironments forming their “context”. The possibility that a given context impacts the response to a given stimulus defines “context-dependency” and it explains large parts of the functional variability of physiopathological and pharmacological stimuli. Currently, there is no framework to analyze and quantify context-dependency over multiple contexts and cellular response outputs. We established an experimental system including a stimulus of interest, applied to an immune cell type in several contexts. We studied the function of OX40 ligand (OX40L) on T helper (Th) cell differentiation, in 4 molecular (Th0, Th1, Th2, and Th17) and 11 dendritic cell (DC) contexts (monocyte-derived DC and cDC2 conditions). We measured 17 Th output cytokines in 302 observations, and developed a statistical modeling strategy to quantify OX40L context-dependency. This revealed highly variable context-dependency, depending on the output cytokine and context type itself. Among molecular contexts, Th2 was the most influential on OX40L function. Among DC contexts, the DC type rather than the activating stimuli was dominant in controlling OX40L context-dependency. This work mathematically formalizes the complex determinants of OX40L functionality, and provides a unique framework to decipher and quantify the context-dependent variability of any biomolecule or drug function.

Measuring context dependency in birdsong using artificial neural networks

Context dependency is a key feature in sequential structures of human language, which requires reference between words far apart in the produced sequence. Assessing how long the past context has an effect on the current status provides crucial information to understand the mechanism for complex sequential behaviors. Birdsongs serve as a representative model for studying the context dependency in sequential signals produced by non-human animals, while previous reports were upper-bounded by methodological limitations. Here, we newly estimated the context dependency in birdsongs in a more scalable way using a modern neural-network-based language model whose accessible context length is sufficiently long. The detected context dependency was beyond the order of traditional Markovian models of birdsong, but was consistent with previous experimental investigations. We also studied the relation between the assumed/auto-detected vocabulary size of birdsong (i.e., fine- vs. coarse-grained syllable classifications) and the context dependency. It turned out that the larger vocabulary (or the more fine-grained classification) is assumed, the shorter context dependency is detected.

It More than Adds Up: Interaction of Antibiotic Mixing and Temperature

Use of antibiotics for the treatment and prevention of bacterial infections in humans, agri- and aquaculture as well as livestock rearing leads to antibiotic pollution of fresh water and these antibiotics have an impact on free-living bacteria. While we know which antibiotics are most common in natural environments such as rivers and streams, there is considerable uncertainty regarding antibiotics’ interactions with one another and the effect of abiotic factors such as temperature. Here, we used an experimental approach to explore the effects of antibiotic identity, concentration, mixing and water temperature on the growth of Pseudomonas fluorescens, a common, ubiquitous bacterium. We exposed P. fluorescens to the four antibiotics most commonly found in surface waters (ciprofloxacin, ofloxacin, sulfamethoxazole and sulfapyridine) and investigated antibiotic interactions for single and mixed treatments at different, field-realistic temperatures. We observed an overall dependence of antibiotic potency on temperature, as temperature increased efficacy of ciprofloxacin and ofloxacin with their EC50 lowered by >75% with a 10 °C temperature increase. Further, we show that mixtures of ciprofloxacin and ofloxacin, despite both belonging to the fluoroquinolone class, exhibit low-temperature-dependent synergistic effects in inhibiting bacterial growth. These findings highlight the context dependency of antibiotic efficacy. They further suggest antibiotic-specific off-target effects that only affect the bacteria once they enter a certain temperature range. This has important implications as freshwater systems already contain multi-drug antibiotic cocktails and are changing temperature due to environmental warming. These factors will interact and affect aquatic food webs, and hence this creates an urgent need to adapt and improve laboratory testing conditions to closer reflect natural environments.

Naming and Indexicality

How do words stand for things? Taking ideas from philosophical semantics and pragmatics, this book offers a unique, detailed, and critical survey of central debates concerning linguistic reference in the twentieth century. It then uses the survey to identify and argue for a novel version of current 'two-dimensional' theories of meaning, which generalise the context-dependency of indexical expressions. The survey highlights the history of tensions between semantic and epistemic constraints on plausible theories of word meaning, from analytic philosophy and modern truth-conditional semantics, to the Referentialist and Externalist revolutions in theories of meaning, to the more recent reconciliatory ambition of two-dimensionalists. It clearly introduces technical semantical notions, theses, and arguments, with easy-to-follow, step-by-step guides. Wide-ranging in its scope, yet offering an accessible route into literature that can seem complex and technical, this will be essential reading for advanced students, and academic researchers in semantics, pragmatics, and philosophy of language.