ALBERT

Beschreibung: 〈p〉Publication date: January 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 128〈/p〉 〈p〉Author(s): J.M. Kranabetter〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Forest floor carbon (C) sequestration has been negatively correlated with manganese (Mn) availability, possibly due to reduced efficacy of Mn-peridoxase enzymes produced by Agaricomycete fungi. I examined a soil C and Mn dataset from a podzolization gradient, along with fungal sporocarp Mn concentrations, to potentially corroborate this finding. An inverse power relationship between soil C and soil Mn content across temperate rainforests was confirmed, which provides further evidence of a Mn bottleneck in C turnover. Average Mn concentrations of saprotrophic sporocarps were greater than those of ectomycorrhizal fungi, and displayed a similar inverse correlation with increasing soil C. The absence or limited effectiveness of select saprotrophic fungi across Mn-depleted forest soils may be one mechanism behind impeded turnover of recalcitrant organic matter.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: January 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 128〈/p〉 〈p〉Author(s): Charles R. Warren〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Phospholipids are isolated from crude lipid extracts by silica solid phase extraction (SPE), but for soils we don't know if phospholipids are the only fatty acid-based lipids present in the polar lipid fraction. Lipids extracted from three soils were fractionated with a silica SPE protocol commonly used for soils, with “neutrals” eluted by chloroform, “glycolipids” eluted by acetone, and “phospholipids” eluted by methanol. Fatty acid-based lipids were identified and quantified by liquid chromatography-mass spectrometry. Phospholipids were recovered in the methanol fraction, but this fraction also included betaine lipids. In two soils the methanol fraction was 3–6% betaine lipid while in one soil betaine lipids accounted for 48% of lipids in the methanol fraction. Clearly the fraction obtained by eluting lipids from silica with methanol is not purely phospholipid but can contain significant amounts of other polar lipids.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: December 2018〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 127〈/p〉 〈p〉Author(s): Aikaterini Efthymiou, Birgit Jensen, Iver Jakobsen〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Biochar (BC) application to soil can potentially replace mineral P fertilizers and its effectiveness as fertilizer can be improved by plant inoculation with the phosphate-solubilizing microorganism 〈em〉Penicillium aculeatum〈/em〉 (Pa). Arbuscular mycorrhiza (AM) fungi are important in plant P nutrition and may possibly act synergistically with Pa to improve the uptake of BC-P. Responses in wheat to inoculation with Pa and the AM fungus 〈em〉Rhizophagus irregularis〈/em〉 were studied in a pot experiment at two levels of BC fertilization. Pots contained a mesh-enclosed, root free compartment with 33P-labelled soil for assessment of the AM contribution to P uptake and for studying if Pa would affect P uptake by the AM fungal hyphae. AM application suppressed wheat growth, albeit AM pathway had a major role in total P uptake at the two lowest P levels (nil or 20 mg BC-P kg〈sup〉−1〈/sup〉 soil). Moreover, AM contribution had similar magnitudes in the presence and absence of Pa. Rhizosphere and bulk soil were actively colonized by Pa, both in the presence and absence of AM. The application of Pa or BC at a low rate increased AM-colonized root lengths. Although this was not translated to increased P uptake by wheat, the results suggest that AM and Pa can be combined without showing antagonistic interactions. However, more work is needed to understand how AM and Pa can be combined to increase plant growth.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 87〈/p〉 〈p〉Author(s): Jimin Xiao, Yanchun Xie, Tammam Tillo, Kaizhu Huang, Yunchao Wei, Jiashi Feng〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Person search in real-world scenarios is a new challenging computer version task with many meaningful applications. The challenge of this task mainly comes from: (1) unavailable bounding boxes for pedestrians and the model needs to search for the person over the whole gallery images; (2) huge variance of visual appearance of a particular person owing to varying poses, lighting conditions, and occlusions. To address these two critical issues in modern person search applications, we propose a novel Individual Aggregation Network (IAN) that can accurately localize persons by learning to minimize intra-person feature variations. IAN is built upon the state-of-the-art object detection framework, i.e., faster R-CNN, so that high-quality region proposals for pedestrians can be produced in an online manner. In addition, to relieve the negative effect caused by varying visual appearances of the same individual, IAN introduces a novel center loss that can increase the intra-class compactness of feature representations. The engaged center loss encourages persons with the same identity to have similar feature characteristics. Extensive experimental results on two benchmarks, i.e., CUHK-SYSU and PRW, well demonstrate the superiority of the proposed model. In particular, IAN achieves 77.23% mAP and 80.45% top-1 accuracy on CUHK-SYSU, which outperform the state-of-the-art by 1.7% and 1.85%, respectively.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Shengcong Chen, Changxing Ding, Minfeng Liu〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Brain tumor segmentation from Magnetic Resonance Imaging scans is vital for both the diagnosis and treatment of brain cancers. It is widely accepted that accurate segmentation depends on multi-level information. However, exiting deep architectures for brain tumor segmentation fail to explicitly encourage the models to learn high-quality hierarchical features. In this paper, we propose a series of approaches to enhance the quality of the learnt hierarchical features. Our contributions incorporate four aspects. First, we extend the popular DeepMedic model to Multi-Level DeepMedic to make use of multi-level information for more accurate segmentation. Second, we propose a novel dual-force training scheme to promote the quality of multi-level features learnt from deep models. It is a general training scheme and can be applied to many exiting architectures, e.g., DeepMedic and U-Net. Third, we design a label distribution-based loss function as an auxiliary classifier to encourage the high-level layers of deep models to learn more abstract information. Finally, we propose a novel Multi-Layer Perceptron-based post-processing approach to refine the prediction results of deep models. Extensive experiments are conducted on two most recent brain tumor segmentation datasets, i.e., BRATS 2017 and BRATS 2015 datasets. Results on the two databases indicate that the proposed approaches consistently promote the segmentation performance of the two popular deep models.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: February 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 129〈/p〉 〈p〉Author(s): Raúl Ochoa-Hueso, Manuel Delgado-Baquerizo, Paul Tuan An King, Merryn Benham, Valentina Arca, Sally A. Power〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Litter decomposition is fundamental for nutrient and carbon (C) cycling, playing a major role in regulating the Earth's climate system. Climate change and fertilization are expected to largely shift litter decomposition rates in terrestrial ecosystems, however, studies contextualizing the relative importance of these major global change drivers versus other key decomposition drivers such as substrate quality and ecosystem type are lacking. Herein, we used two independent field experiments in an Eastern Australian grassland (Experiment 1) and a forest (Experiment 2) to evaluate the role of (i) litter quality, (ii) nutrient addition (N, P and K in full factorial combination; Experiment 1), and (iii) a combination of N addition and irrigation (Experiment 2) in litter decomposition, substrate-induced respiration and microbial abundance. Regardless of experimental treatments, forest soils decomposed litter between 2 and 5 times faster than grassland soils. This was principally controlled by the greater ability of forest microbes to respire C-based substrates and, ultimately, by soil N availability. The experimental treatments accounted for only relatively small differences in our measured variables, ranging from 10 to 15% in the case of the irrigation-by-N-addition forest experiment to almost negligible in most of the grassland nutrient addition plots. In the latter experiment, decomposition and soil activity responses were associated with either K addition or interactions between K and other nutrients, suggesting a key role for this often-neglected soil nutrient in controlling litter decomposition. Our study provides evidence that while nutrient enrichment and/or irrigation have the potential to affect litter decomposition rates in grassland and forest ecosystems, land use change that results in loss or gain of forested area is likely to exert a much greater impact than these other two drivers.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: February 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 129〈/p〉 〈p〉Author(s): Esther Guillot, Philippe Hinsinger, Lydie Dufour, Jacques Roy, Isabelle Bertrand〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Soil microbial communities in Mediterranean agroecosystems experience long drought periods that are typically combined with heat and frequently interspersed with rapid rewetting events. Agroforestry systems are of growing interest and viewed as possible alternative to conventional cropping systems in the context of climate change. Our aim was to evaluate the resistance and resilience of soil microbial communities against drought with or without heat stress at different distances from the tree row in an agroforestry system as compared to a conventional cropping system. Soils were sampled at several distances from the tree row in a 21-year-old walnut agroforestry system and a contiguous conventional crop in Southern France. We simulated two cycles of drying-rewetting under controlled conditions and applied three distinct treatments: control (without stress), drought and drought combined with heat stress. We monitored microbial respiration over the incubation period. The inorganic N and microbial biomass C, N and P contents (MBC, MBN and MBP) were assessed during the drying period (resistance), just after rewetting and at the end of the experiment (resilience), while bacterial and fungal abundances were measured at the end of the resistance period. We demonstrated that an agroforestry system can induce spatial heterogeneity in soil microbial biomass and functions under control conditions. Microbial biomass and activity, soil organic matter (SOM) and mineral N levels increased on the tree row. This spatial heterogeneity pattern was preserved for soil microbial response to drought combined or not with heat. Microorganisms sampled in the middle of the interrow or in the conventional crop exhibited highest biomass resistance and lowest resilience when facing combined drought and heat stress. Soil microbial biomass resistance and resilience were similar whatever the spatial position when microorganisms had to deal with drought stress only. Our findings suggested that despite higher SOM content, microbial biomass and activity at and near the tree row, the legacy effect of the tree row did not lead to higher ecological stability under stressful climatic conditions. We also demonstrated that soil microorganisms can considerably change their stoichiometry depending on the stress treatment. Soil microorganisms showed elevated MBC:MBN, MBC:MBP and variable MBN:MBP during the resistance period. A high stoichiometric flexibility of microorganisms was observed when exposed to drought stress only, while stoichiometric changes were irreversible when exposed to combined drought and heat stress.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: February 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 129〈/p〉 〈p〉Author(s): Yufang Lu, Xiaonan Zhang, Jiafeng Jiang, Herbert J. Kronzucker, Weishou Shen, Weiming Shi〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉The application of biological nitrification inhibitors (BNIs) is considered an important new strategy to mitigate nitrogen losses from agricultural soils. 1,9-decanediol was recently identified as a new BNI in rice root exudates and was shown to inhibit nitrification in bioassays using 〈em〉Nitrosomonas〈/em〉. However, the effect of this compound on nitrification and ammonia oxidizers in soils remained unknown. In this study, three typical agriculture soils were collected to investigate the impact of 1,9-decanediol on nitrification and ammonia oxidizers in a 14-day microcosm incubation. High doses of 1,9-decanediol showed strong soil nitrification inhibition in all three agricultural soils, with the highest inhibition of 58.1% achieved in the acidic red soil, 37.0% in the alkaline fluvo-aquic soil, and 35.7% in the neutral paddy soil following 14 days of incubation. Moreover, the inhibition of 1,9-decanediol was superior to the widely used synthetic nitrification inhibitor, dicyandiamide (DCD) and two other BNIs, methyl 3-(4-hydroxyphenyl) propionate (MHPP) and α-linolenic acid (LN), in all three soils. The abundance of ammonia-oxidizing bacteria (AOB) and ammonia-oxidizing archaea (AOA) was significantly inhibited by 1,9-decanediol addition across the three soils. All AOB sequences fell within the 〈em〉Nitrosospira〈/em〉 group, and the dominant AOA sequences belonged to the 〈em〉Nitrososphaera〈/em〉 cluster in all three soils. Changes in the community composition of AOB were more pronounced than AOA after the application of 1,9-decanediol. The AOB community structure shifted from 〈em〉Nitrosospira〈/em〉 cluster 2 and cluster 3a toward 〈em〉Nitrosospira〈/em〉 clusters 8a and 8b. As for AOA, no significant impact on the proportion of the dominant 〈em〉Nitrososphaera〈/em〉 cluster was observed in the fluvo-aquic soil and paddy soil while only the 〈em〉Nitrosopumilus〈/em〉 cluster decreased in the red soil. 1,9-decanediol could also significantly reduce soil N〈sub〉2〈/sub〉O emissions, especially in acidic red soil. Our results provide evidence that 1,9-decanediol is capable of suppressing nitrification in agricultural soils through impeding both AOA and AOB rather than affecting soil NH〈sub〉4〈/sub〉〈sup〉+〈/sup〉 availability. 1,9-decanediol holds promise as an effective biological nitrification inhibitor for soil ammonia-oxidizing bacteria and archaea.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): Andrew R. Zimmerman, Lei Ouyang〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Due to expected changes in fire frequency and the potential of using pyrolyzed biomass (biochar) amendments to increase soil C storage, there is a need for better ability to predict pyrogenic C (pyC) longevity in soil and its effects on native soil C stability. However, C mineralization from biochar/soil mixtures has been shown to vary greatly and both ‘positive’ and ‘negative’ priming (increased and decreased mineralization of native C, respectively) following biochar amendments have been observed. To better understand the interactions that influence mineralization of pyC and native soil C, bagasse (sugar cane residues) and bagasse biochar pyrolyzed at 300 and 650 °C were incubated in sand over 144 d with soil microbes and dissolved organic matter (DOM) substrates of high and low reactivity (sucrose and humic acid: HA, respectively). Mineralization of particulate and dissolved C was quantified based upon the distinct C isotopic signature of CO〈sub〉2〈/sub〉 evolved from each source. Negative priming of bagasse-C mineralization by sucrose (−9.3% cumulative C mineralized) and pyC mineralization by HA (−29 and −68% for low and high temperature biochar, respectively) pointed to the mechanism of ‘〈em〉substrate switching〈/em〉’, i.e. cases in which added DOM was of greater or similar lability to the particulate OM present. In contrast, positive priming of bagasse mineralization by HA (+77%) and pyC mineralization by sucrose (+271 and 614% for low and high temperature biochar, respectively), was attributed to the mechanisms of 〈em〉soil conditioning〈/em〉 (creation of an environment more favourable to microbial growth) and 〈em〉co-metabolism〈/em〉, respectively. Inversely, presence of all the particulates enhanced the mineralization of sucrose (by 8, 58 and 91% for bagasse and low and high temperature biochar, respectively), suggesting a 〈em〉soil conditioning〈/em〉 mechanism. In contrast, the biochars had little effect on HA mineralization, likely because of their similar inherent stability and chemistry. These results show that DOM and pyC mineralization in soil is interactive and varies with OM type. Furthermore, the priming observed could be attributed to different mechanisms in different cases, the long term effect of which would likely be greater soil C sequestration than predicted by simple degradation models.〈/p〉〈/div〉 〈/div〉 〈h5〉Graphical abstract〈/h5〉 〈div〉〈p〉〈figure〉〈img src="https://ars.els-cdn.com/content/image/1-s2.0-S0038071718304218-fx1.jpg" width="404" alt="Image 1" title="Image 1"〉〈/figure〉〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Peizhen Bai, Yan Ge, Fangling Liu, Haiping Lu〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉In recommender systems, the classical matrix factorization model for collaborative filtering only considers joint interactions between users and items. In contrast, context-aware recommender systems (CARS) use contexts to improve recommendation performance. Some early CARS models treat user, item and context equally, unable to capture contextual impact accurately. More recent models perform context operations on users and items separately, leading to “double-counting” of contextual information. This paper proposes a new model, Joint Interaction with Context Operation (JICO), to integrate the joint interaction model with the context operation model, via two layers. The joint interaction layer models interactions between users and items via an interaction tensor. The context operation layer captures contextual information via a contextual operating tensor. We evaluate JICO on four datasets and conduct novel studies, including varying contextual influence and time split recommendation. JICO consistently outperforms competing methods, while providing many useful insights to assist further analysis.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 131〈/p〉 〈p〉Author(s): Emma L. Aronson, Michael L. Goulden, Steven D. Allison〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉 〈p〉Climate and regional air quality models predict that Southern California will experience longer and more severe droughts, and possibly wetter, more intense storms and changing nitrogen (N) deposition. We investigated how the three major soil greenhouse gas (GHG) fluxes respond to 4–6 years of exposure to a full-factorial experiment of reduced and augmented precipitation crossed with increased N in a semi-arid grassland in Irvine, CA, USA. The mean emission fluxes across all treatments were 249.8 mg CO〈sub〉2〈/sub〉 m〈sup〉−2〈/sup〉 h〈sup〉−1〈/sup〉, -16.41 μg CH〈sub〉4〈/sub〉 m〈sup〉−2〈/sup〉 h〈sup〉−1〈/sup〉, and 2.24 μg N〈sub〉2〈/sub〉O m〈sup〉−2〈/sup〉 h〈sup〉−1〈/sup〉. Added N plots released 3.5 times more N〈sub〉2〈/sub〉O than ambient N plots, and N treatment and soil moisture interacted, such that volumetric soil moisture in added N plots correlated positively with N〈sub〉2〈/sub〉O release. Soil moisture, which was higher in the added water plots, correlated positively with respiration. CH〈sub〉4〈/sub〉 consumption increased with soil moisture in the drought treatment, an opposite trend to that observed in most other studies.〈/p〉 〈p〉Our data suggest that CH〈sub〉4〈/sub〉 consumption, N〈sub〉2〈/sub〉O production, and soil respiration will decline if Southern California grasslands experience more frequent and extreme droughts. However, when drought is followed by high rainfall, the additional moisture will likely increase CH〈sub〉4〈/sub〉 consumption and N〈sub〉2〈/sub〉O release in periodic pulses. Overall, climatic shifts in this ecosystem may lead to a decrease in overall soil GHG emissions to the atmosphere. However, increased N deposition to Southern California will likely lead to increased N〈sub〉2〈/sub〉O release and a shift in the dominant N loss pathway toward gaseous release of N. If N deposition continues to increase along with severity and duration of drought, our data predict a decrease in global warming potential (GWP) of 17.2% from this ecosystem.〈/p〉 〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): Liang Chen, Wenhua Xiang, Huili Wu, Shuai Ouyang, Bo Zhou, Yelin Zeng, Yongliang Chen, Yakov Kuzyakov〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Plant interactions and feedbacks with soil microorganisms play an important role in sustaining the functions and stability of terrestrial ecosystems, yet the effects of tree species diversity on soil microbial community in forest ecosystems are still not well understood. Here, we examined the effects of tree species richness (1–12 species) and the presence of certain influential tree species (sampling effect) on soil bacterial and fungal communities in Chinese subtropical forests, using high-throughput Illumina sequencing for microbial identification. We observed that beta rather than alpha diversities of tree species and soil microorganisms were strong coupled. Multivariate regression and redundancy analyses revealed that the effects of tree species identity dominated over tree species richness on the diversity and composition of bacterial and fungal communities in both organic and top mineral soil horizons. Soil pH, nutrients and topography were always identified as significant predictors in the best multivariate models. Tree species have stronger effect on fungi than bacteria in organic soil, and on ectomycorrhizal fungi than saprotrophic fungi in mineral topsoil. Concluding, tree species identity, along with abiotic soil and topographical conditions, were more important factors determining the soil microbial communities in subtropical forests than tree diversity 〈em〉per se〈/em〉.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Mingyuan Jiu, Hichem Sahbi〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Deep kernel learning aims at designing nonlinear combinations of multiple standard elementary kernels by training deep networks. This scheme has proven to be effective, but intractable when handling large-scale datasets especially when the depth of the trained networks increases; indeed, the complexity of evaluating these networks scales quadratically w.r.t. the size of training data and linearly w.r.t. the depth of the trained networks. In this paper, we address the issue of efficient computation in Deep Kernel Networks (DKNs) by designing effective maps in the underlying Reproducing Kernel Hilbert Spaces (RKHS). Given a pretrained DKN, our method builds its associated Deep Map Network (DMN) whose inner product approximates the original network while being far more efficient. The design principle of our method is greedy and achieved layer-wise, by finding maps that approximate DKNs at different (input, intermediate and output) layers. This design also considers an extra fine-tuning step based on unsupervised learning, that further enhances the generalization ability of the trained DMNs. When plugged into SVMs, these DMNs turn out to be as accurate as the underlying DKNs while being at least an order of magnitude faster on large-scale datasets, as shown through extensive experiments on the challenging ImageCLEF, COREL5k benchmarks and the Banana dataset.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Yu Zhang, Han Zhang, Xiaobo Chen, Mingxia Liu, Xiaofeng Zhu, Seong-Whan Lee, Dinggang Shen〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Sparse representation-based brain functional network modeling often results in large inter-subject variability in the network structure. This could reduce the statistical power in group comparison, or even deteriorate the generalization capability of the individualized diagnosis of brain diseases. Although group sparse representation (GSR) can alleviate such a limitation by increasing network similarity across subjects, it could, in turn, fail in providing satisfactory separability between the subjects from different groups (e.g., patients vs. controls). In this study, we propose to integrate individual functional connectivity (FC) information into the GSR-based network construction framework to achieve higher between-group separability while maintaining the merit of within-group consistency. Our method was based on an observation that the subjects from the same group have generally more similar FC patterns than those from different groups. To this end, we propose our new method, namely “strength and similarity guided GSR (SSGSR)”, which exploits both BOLD signal temporal correlation-based “low-order” FC (LOFC) and inter-subject LOFC-profile similarity-based “high-order” FC (HOFC) as two priors to jointly guide the GSR-based network modeling. Extensive experimental comparisons are carried out, with the rs-fMRI data from mild cognitive impairment (MCI) subjects and healthy controls, between the proposed algorithm and other state-of-the-art brain network modeling approaches. Individualized MCI identification results show that our method could achieve a balance between the individually consistent brain functional network construction and the adequately maintained inter-group brain functional network distinctions, thus leading to a more accurate classification result. Our method also provides a promising and generalized solution for the future connectome-based individualized diagnosis of brain disease.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): Denis Curtin, Mike H. Beare, Weiwen Qiu, Craig S. Tregurtha〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Phosphorus (P) is central to the productivity of grass-clover pastoral farming systems. Fertiliser P promotes the growth of clover, which provides an input of fixed N to complement that supplied from the soil (primarily mineralised N). If pasture productivity is limited by P availability, organic matter returns to the soil in excreta and plant residues will decline, which in turn may reduce the supply of N by mineralisation. We examined the effect of long-term P application to grass-clover pasture on mineralisation of N (and C). Net N mineralisation was measured in a 14-week aerobic incubation (25 °C; soil maintained at field capacity) using soils (0–15 cm depth) from a long-term (1952–2016) trial set up to quantify effects of single superphosphate (0, 188, or 376 kg〈sup〉−1〈/sup〉 year〈sup〉−1〈/sup〉) on productivity of irrigated, sheep-grazed pasture (no fertiliser N was applied during the trial). Although P fertilisation had only a small effect (∼10% increase) on total soil N, net N mineralisation was substantially increased (N mineralised from fertilised soil in 14 weeks was ∼1.6 times that from the unfertilised Control). In contrast, mineralisation of C was slightly greater in the Control than in fertilised soil. Nitrogen mineralisation exhibited a Mitscherlich-type relationship with available soil P, measured using the Olsen test; near-maximum mineralisation was observed at an Olsen test value of ∼10 mg P kg〈sup〉−1〈/sup〉 soil. Annual in-field N mineralisation was estimated by modifying the laboratory-measured “basal” mineralisation values using temperature and moisture adjustment factors (soil temperature and moisture data were acquired from an adjacent, irrigated trial). The results confirmed that N mineralisation was the predominant source of available N in the unfertilised Control (∼160 kg ha〈sup〉−1〈/sup〉 yr〈sup〉−1〈/sup〉 vs ∼30 kg ha〈sup〉−1〈/sup〉 yr〈sup〉−1〈/sup〉 of fixed N) and that it was an important source of N for the additional dry matter grown in response to P application (N uptake increased by ∼200 kg ha〈sup〉−1〈/sup〉 yr〈sup〉−1〈/sup〉 in response to P fertilisation vs an increase in mineralisation of ∼100 kg ha〈sup〉−1〈/sup〉 yr〈sup〉−1〈/sup〉). The increase in net N mineralisation in fertilised soil was partly because immobilisation of N was less than in the unfertilised Control.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Haishan Ye, Guangzeng Xie, Luo Luo, Zhihua Zhang〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Optimization is an important issue in machine learning because many machine learning models are reformulated as optimization problems. Different kinds of machine learning algorithms mainly focus on minimizing their empirical loss like deep learning, logistic regression, and support vector machine. Because data is explosively growing, it is challenging to deal with a large-scale optimization problem. Recently, stochastic second-order methods have emerged to attract much attention due to their efficiency in each iteration. These methods show good performance on training machine learning algorithms like logistic regression and support vector machine. However, the computational complexity of existing stochastic second-order methods heavily depends on the condition number of the Hessian. In this paper, we propose a new Newton-like method called 〈em〉Preconditioned Newton Conjugate Gradient with Sketched Hessian〈/em〉 (PNCG). The runtime complexity of PNCG is at most 〈em〉logarithmic〈/em〉 in the condition number of the Hessian. PNCG exhibits advantages over existing subsampled Newton methods especially when the Hessian matrix in question is ill-conditioned. We also show that our method has good performance on training machine learning algorithm empirically. The results show consistent improvements in computational efficiency.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): Eric R. Johnston, Minjae Kim, Janet K. Hatt, Jana R. Phillips, Qiuming Yao, Yang Song, Terry C. Hazen, Melanie A. Mayes, Konstantinos T. Konstantinidis〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Tropical ecosystems are an important sink for atmospheric CO〈sub〉2〈/sub〉; however, plant growth is restricted by phosphorus (P) availability. Although soil microbiota facilitate organic P turnover and inorganic P mobilization, their role in carbon-phosphorus coupled processes remains poorly understood. To advance this topic, soils collected from four sites representing highly weathered tropical soils in the El Yunque National Forest, Puerto Rico were incubated with exogenous PO〈sub〉4〈/sub〉〈sup〉3−〈/sup〉 under controlled laboratory conditions. P amendment increased CO〈sub〉2〈/sub〉 respiration by 14–23% relative to control incubations for soils sampled from all but the site with the greatest total and bioavailable soil P. Metatranscriptomics revealed an increase in the relative transcription of genes involved in cell growth and uptake of other nutrients in response to P amendment. A new methodology to normalize gene expression by population-level relative (DNA) abundance revealed that the pattern of increased transcription of cell growth and division genes with P amendment was community-wide. Soil communities responsive to P amendment possessed a greater relative abundance of α-glucosyl polysaccharide biosynthesis genes, suggestive of enhanced C storage under P-limiting conditions. Phosphorylase genes governing the degradation of α-glucosyl polysaccharides were also more abundant and increased in relative transcription with P amendment, indicating a shift from energy storage towards growth. Conversely, microbial communities in soils nonresponsive to P amendment were found to have metabolisms tuned for the phosphorolysis of labile plant-derived substrates, such as β-glucosyl polysaccharides. Collectively, our results provided quantitative estimates of increased soil respiration upon alleviation of P constraints and elucidated several underlying ecological and molecular mechanisms involved in this response.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Pattathal V. Arun, Ittai Herrmann, Krishna M. Budhiraju, Arnon Karnieli〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Spatial resolution enhancement is a pre-requisite for integrating unmanned aerial vehicle (UAV) datasets with the data from other sources. However, the mobility of UAV platforms, along with radiometric and atmospheric distortions, makes the task difficult. In this paper, various convolutional neural network (CNN) architectures are explored for resolving the issues related to sub-pixel classification and super-resolution of drone-derived datasets. The main contributions of this work are: 1) network-inversion based architectures for super-resolution and sub-pixel mapping of drone-derived images taking into account their spectral-spatial characteristics and the distortions prevalent in them 2) a feature-guided transformation for regularizing the inversion problem 3) loss functions for improving the spectral fidelity and inter-label compatibility of coarser to finer-scale mapping 4) use of multi-size kernel units for avoiding over-fitting. The proposed approach is the first of its kind in using neural network inversion for super-resolution and sub-pixel mapping. Experiments indicate that the proposed super-resolution approach gives better results in comparison with the sparse-code based approaches which generally result in corrupted dictionaries and sparse codes for multispectral aerial images. Also, the proposed use of neural network inversion, for projecting spatial affinities to sub-pixel maps, facilitates the consideration of coarser-scale texture and color information in modeling the finer-scale spatial-correlation. The simultaneous consideration of spectral bands, as proposed in this study, gives better super-resolution results when compared to the individual band enhancements. The proposed use of different data-augmentation strategies, for emulating the distortions, improves the generalization capability of the framework. Sensitivity of the proposed super-resolution and sub-pixel mapping frameworks with regard to the network parameters is thoroughly analyzed. The experiments over various standard datasets as well as those collected from known locations indicate that the proposed frameworks perform better when compared to the prominent published approaches.〈/p〉〈/div〉 〈h5〉Graphical abstract〈/h5〉 〈div〉〈p〉〈figure〉〈img src="https://ars.els-cdn.com/content/image/1-s2.0-S0031320318304217-fx1.jpg" width="221" alt="Image, graphical abstract" title="Image, graphical abstract"〉〈/figure〉〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): Gangsheng Wang, Wenjuan Huang, Melanie A. Mayes, Xiaodong Liu, Deqiang Zhang, Qianmei Zhang, Tianfeng Han, Guoyi Zhou〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Knowledge of microbial mechanisms is critical to understand Earth's biogeochemical cycle under climate and environmental changes. However, large uncertainties remain in model simulations and predictions due to the lack of explicit parameterization of microbial data and few applications beyond the laboratory. In addition, most experimental and modeling studies of warming-induced changes in soil carbon (C) focus on temperature sensitivity, neglecting concomitant effects of changes in soil moisture. Soil microbes are sensitive to moisture, and their responses can dramatically impact soil biogeochemical cycles. Here we represent microbial and enzymatic functions in response to changes in moisture in the Microbial-ENzyme Decomposition (MEND) model. Through modeling with long-term field observations from subtropical forests, we demonstrate that parameterization with microbial data in addition to respiration fluxes greatly increases confidence in model simulations. We further employ the calibrated model to simulate the responses of soil organic C (SOC) under multiple environmental change scenarios. The model shows significant increases in SOC in response to decreasing soil moisture and only minor changes in SOC in response to increasing soil temperature. Increasing litter inputs also cause a significant increase in SOC in the pine forest, whereas an insignificant negative effect is simulated in the broadleaf forest. We also demonstrate the co-metabolism mechanism for the priming effects, i.e., more labile inputs to soil could stimulate microbial and enzymatic growth and activity. Our study provides strong evidence of microbial control over soil C decomposition and suggests the future trajectory of soil C may be more responsive to changes in soil moisture than temperature, particularly in tropical and subtropical environments.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 131〈/p〉 〈p〉Author(s): Abdallah Awad, Andrzej Majcherczyk, Peter Schall, Kristina Schröter, Ingo Schöning, Marion Schrumpf, Martin Ehbrecht, Steffen Boch, Tiemo Kahl, Jürgen Bauhus, Dominik Seidel, Christian Ammer, Markus Fischer, Ursula Kües, Rodica Pena〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Functionally, ectomycorrhizal (ECM) and saprotrophic (SAP) fungi belong to different guilds, and they play contrasting roles in forest ecosystem C-cycling. SAP fungi acquire C by degrading the soil organic material, which precipitates massive CO〈sub〉2〈/sub〉 release, whereas, as plant symbionts, ECM fungi receive C from plants representing a channel of recently assimilated C to the soil. In this study, we aim to measure the amounts and identify the drivers of ECM and SAP fungal biomass in temperate forest topsoil. To this end, we measured ECM and SAP fungal biomass in mineral topsoils (0–12 cm depth) of different forest types (pure European beech, pure conifers, and mixed European beech with other broadleaf trees or conifers) in a range of about 800 km across Germany; moreover, we conducted multi-model inference analyses using variables for forest and vegetation, nutritive resources from soil and roots, and soil conditions as potential drivers of fungal biomass. Total fungal biomass ranged from 2.4 ± 0.3 mg g〈sup〉−1〈/sup〉 (soil dry weight) in pure European beech to 5.2 ± 0.8 mg g〈sup〉−1〈/sup〉 in pure conifer forests. Forest type, particularly the conifer presence, had a strong effect on SAP biomass, which ranged from a mean value of 1.5 ± 0.1 mg g〈sup〉−1〈/sup〉 in broadleaf to 3.3 ± 0.6 mg g〈sup〉−1〈/sup〉 in conifer forests. The European beech forests had the lowest ECM fungal biomass (1.1 ± 0.3 mg g〈sup〉−1〈/sup〉), but in mixtures with other broadleaf species, ECM biomass had the highest value (2.3 ± 0.2 mg g〈sup〉−1〈/sup〉) among other forest types. Resources from soil and roots such as N and C concentrations or C:N ratios were the most influential variables for both SAP and ECM biomass. Furthermore, SAP biomass were driven by factors related to forest structure and vegetation, whereas ECM biomass was mainly influenced by factors related to soil conditions, such as soil temperature, moisture, and pH. Our results show that we need to consider a complex of factors differentially affecting biomass of soil fungal functional groups and highlight the potential of forest management to control forest C-storage and the consequences of changes in soil fungal biomass.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Chao Gou, Hui Zhang, Kunfeng Wang, Fei-Yue Wang, Qiang Ji〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Image-based pupil detection, which aims to find the pupil location in an image, has been an active research topic in computer vision community. Learning-based approaches can achieve preferable results given large amounts of training data with eye center annotations. However, there are limited publicly available datasets with accurate eye center annotations and it is unreliable and time-consuming for manually labeling large amounts of training data. In this paper, inspired by learning from synthetic data in Parallel Vision framework, we introduce a step of parallel imaging built upon Generative Adversarial Networks (GANs) to generate adversarial synthetic images. In particular, we refine the synthetic eye images by the improved SimGAN using adversarial training scheme. For the computational experiments, we further propose a coarse-to-fine pupil detection framework based on shape augmented cascade regression models learning from the adversarial synthetic images. Experiments on benchmark databases of BioID, GI4E, and LFW show that the proposed work performs significantly better over other state-of-the-art methods by leveraging the power of cascade regression and adversarial image synthesis.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: May 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 89〈/p〉 〈p〉Author(s): Chen-Lin Zhang, Jianxin Wu〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Nowadays, Convolutional Neural Network (CNN) has achieved great success in various computer vision tasks. However, in classic CNN models, convolution and fully connected (FC) layers just perform linear transformations to their inputs. Non-linearity is often added by activation and pooling layers. It is natural to explore and extend convolution and FC layers non-linearly with affordable costs. In this paper, we first investigate the power mean function, which is proved effective and efficient in SVM kernel learning. Then, we investigate the power mean kernel, which is a non-linear kernel having linear computational complexity with the asymmetric kernel approximation function. Motivated by this scalable kernel, we propose Power Mean Transformation, which nonlinearizes both convolution and FC layers. It only needs a small modification on current CNNs, and improves the performance with a negligible increase of model size and running time. Experiments on various tasks show that Power Mean Transformation can improve classification accuracy, bring generalization ability and add different non-linearity to CNN models. Large performance gain on tiny models shows that Power Mean Transformation is especially effective in resource restricted deep learning scenarios like mobile applications. Finally, we add visualization experiments to illustrate why Power Mean Transformation works.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 131〈/p〉 〈p〉Author(s): Adrian Ho, Hyo Jung Lee, Max Reumer, Marion Meima-Franke, Ciska Raaijmakers, Hans Zweers, Wietse de Boer, Wim H. Van der Putten, Paul L.E. Bodelier〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Aerobic oxidation of methane at (circum-)atmospheric concentrations (〈40 ppm〈sub〉v〈/sub〉) has long been assumed to be catalyzed by the as-yet-uncultured high-affinity methanotrophs in well-aerated, non-wetland (upland) soils, the only known biological methane sink globally. Although the low-affinity canonical methanotrophs with cultured representatives have been detected along with the high-affinity ones, their role as a methane sink in upland soils remains enigmatic. Here, we show that canonical methanotrophs can contribute to (circum-)atmospheric methane uptake in agricultural soils. We performed a stable-isotope 〈sup〉13〈/sup〉C〈img src="https://sdfestaticassets-eu-west-1.sciencedirectassets.com/shared-assets/16/entities/sbnd"〉CH〈sub〉4〈/sub〉 labelling incubation in the presence and absence of bio-based residues that were added to the soil to track the flow of methane. Residue amendment transiently stimulated methane uptake rate (〈50 days). Soil methane uptake was sustained throughout the incubation (130 days), concomitant to the enrichment of 〈sup〉13〈/sup〉C〈img src="https://sdfestaticassets-eu-west-1.sciencedirectassets.com/shared-assets/16/entities/sbnd"〉CO〈sub〉2〈/sub〉. The 〈sup〉13〈/sup〉C-enriched phospholipid fatty acids (PLFAs) were distinct in both soils, irrespective of amendments, and were unambiguously assigned almost exclusively to canonical alphaproteobacterial methanotrophs with cultured representatives. 16S rRNA and 〈em〉pmoA〈/em〉 gene sequence analyses revealed that the as-yet-uncultured high-affinity methanotrophs were virtually absent in these soils. The stable-isotope labelling approach allowed to attribute soil methane uptake to canonical methanotrophs, whereas these were not expected to consume (circum-)atmospheric methane. Our findings thus revealed an overlooked reservoir of high-affinity methane-oxidizers represented by the canonical methanotrophs in agriculture-impacted upland soils. Given that upland agricultural soils have been thought to marginally or do not contribute to atmospheric methane consumption due to the vulnerability of the high-affinity methanotrophs, our findings suggest a thorough revisiting of the contribution of agricultural soils, and the role of agricultural management to mitigation of climate change.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: February 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 129〈/p〉 〈p〉Author(s): 〈/p〉

Beschreibung: 〈p〉Publication date: May 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 89〈/p〉 〈p〉Author(s): Kewei Tang, Zhixun Su, Yang Liu, Wei Jiang, Jie Zhang, Xiyan Sun〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Spectral-clustering based methods have recently attracted considerable attention in the field of subspace segmentation. The approximately block-diagonal graphs achieved by this kind of methods usually contain some noise, i.e., nonzero elements in the off-diagonal region, due to outlier contamination or complex intrinsic structure of the dataset. In the experiment of most previous work, the number of the subspaces is often no more than 10. In this situation, this kind of noise almost has no influence on the segmentation results. However, the segmentation performance could be negatively affected by the noise when the number of subspaces is large, which is quite common in the real-world applications. In this paper, we address the problem of LSN subspace segmentation, i.e., large subspace number subspace segmentation. We first show that the approximately block-diagonal graph with the smaller difference in its diagonal blocks will be more robust to the off-diagonal noise mentioned above. Then, by using the infinity norm to control the bound of the difference in the diagonal blocks, we propose infinity norm minimization for LSN subspace segmentation. Experimental results demonstrate the effectiveness of our method.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): Chantal Koechli, Ashley N. Campbell, Charles Pepe-Ranney, Daniel H. Buckley〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Soils represent one of the largest and most active pools of C in the biosphere, and soil respiration represents a major component of global C flux. Fungi are essential to soil carbon cycling due to their propensity for decomposing organic polymers such as cellulose. We performed high throughput sequencing enabled stable isotope probing (HTS-SIP) with 〈sup〉13〈/sup〉C-cellulose to characterize the dynamics of fungi and bacteria during cellulose degradation in an agricultural soil. A total of 1900 fungal taxa were observed and 190 of these assimilated 〈sup〉13〈/sup〉C-cellulose during a 30-day incubation. A majority of 〈sup〉13〈/sup〉C-labeled fungi belonged to 〈em〉Ascomycota〈/em〉, 〈em〉Basidiomycota〈/em〉, and 〈em〉Mucoromycota〈/em〉. However, most 〈sup〉13〈/sup〉C-labeled fungi could not be annotated at the species (71%, 〈em〉n〈/em〉 = 134), or genus (49%, 〈em〉n〈/em〉 = 93) level. 〈em〉Mucoromycota〈/em〉 were 〈sup〉13〈/sup〉C-labeled early, and by day 3 the most abundant 〈sup〉13〈/sup〉C-labeled organism belonged to 〈em〉Mortierella〈/em〉. In contrast, 〈sup〉13〈/sup〉C-labeled 〈em〉Ascomycota〈/em〉 increased in diversity through day 14 and their relative abundance comprised more than 40% of fungal ITS sequences by day 30. These results show that: 〈em〉i〈/em〉) the majority of fungal taxa that assimilated 〈sup〉13〈/sup〉C from 〈sup〉13〈/sup〉C-cellulose are uncultivated and poorly characterized, 〈em〉ii〈/em〉) the beta-diversity of 〈sup〉13〈/sup〉C-labeled fungi changed significantly over time during cellulose degradation, 〈em〉iii〈/em〉) a relatively small number of the 〈sup〉13〈/sup〉C-labeled taxa dominated the community response to cellulose, and 〈em〉iv〈/em〉) fungi incorporated cellulose into DNA more rapidly and in greater numbers than did bacteria. These results show that fungi in a tilled agricultural field respond rapidly to new cellulose inputs, exhibiting complex temporal dynamics that likely drive carbon flow into diverse taxa within the soil community.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): Katerina Papp, Bruce A. Hungate, Egbert Schwartz〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉 〈p〉RNA is considered to be a short-lived molecule, indicative of cellular metabolic activity, whereas DNA is thought to turn over more slowly because living cells do not always grow and divide. To explore differences in the rates of synthesis of these nucleic acids, we used H〈sub〉2〈/sub〉〈sup〉18〈/sup〉O quantitative stable isotope probing (qSIP) to measure the incorporation of 〈sup〉18〈/sup〉O into 16S rRNA, the 16S rDNA, 〈em〉amoA〈/em〉 mRNA and the 〈em〉amoA〈/em〉 gene of soil Thaumarchaeota.〈/p〉 〈p〉Incorporation of 〈sup〉18〈/sup〉O into the thaumarchaeal 〈em〉amoA〈/em〉 mRNA pool was faster than into the 16S rRNA pool, suggesting that Thaumarchaea were metabolically active while using rRNA molecules that were likely synthetized prior to H〈sub〉2〈/sub〉〈sup〉18〈/sup〉O addition. Assimilation rates of 〈sup〉18〈/sup〉O into 16S rDNA and 〈em〉amoA〈/em〉 genes were similar, which was expected because both genes are present in the same thaumarchaeal genome. The Thaumarchaea had significantly higher rRNA to rDNA ratios than bacteria, though the 〈sup〉18〈/sup〉O isotopic signature of thaumarchaeal rRNA was lower than that of bacterial rRNA, further suggesting preservation of old non-labeled rRNA. Through qSIP of soil with H〈sub〉2〈/sub〉〈sup〉18〈/sup〉O, we showed that 〈sup〉18〈/sup〉O incorporation into thaumarchaeal nucleic acids was generally low, indicating slower turnover rates compared to bacteria, and potentially suggesting thaumarchaeal capability for preservation and efficient reuse of biomolecules.〈/p〉 〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: May 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 89〈/p〉 〈p〉Author(s): Eliezer Flores, Maciel Zortea, Jacob Scharcanski〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Land-use classification in very high spatial resolution images is critical in the remote sensing field. Consequently, remarkable efforts have been conducted towards developing increasingly accurate approaches for this task. In recent years, deep learning has emerged as a dominant paradigm for machine learning, and methodologies based on deep convolutional neural networks have received particular attention from the remote sensing community. These methods typically utilize transfer learning and/or data augmentation to accommodate a small number of labeled images in the publicly available datasets in this field. However, they typically require powerful computers and/or a long time for training. In this work, we propose a simple and novel method for land-use classification in very high spatial resolution images, which efficiently combines transfer learning with a sparse representation. Specifically, the proposed method performs the classification of land-use scenes using a modified version of the well-known sparse representation-based classification method. While this method directly uses the training images to form dictionaries, which are employed to classify test images, our method utilizes a pre-trained deep convolutional neural network and the Gaussian mixture model to generate more robust and compact “dictionaries of deep features.” The effectiveness of the proposed method was evaluated on two publicly available datasets: UC Merced and Brazilian Cerrado–Savana. The experimental results suggest that our method can potentially outperform state-of-the-art techniques for land-use classification in very high spatial resolution images.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Shima Kashef, Hossein Nezamabadi-pour〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉In multi-label data, each instance is associated with a set of labels, instead of one label. Similar to single-label data, feature selection plays an important role in improving classification performance. In multi-label classification, each class label might be specified by some particular characteristics of its own which are called label-specific features. In this paper, a fast accurate filter-based feature selection method is exclusively designed for multi-label datasets to find label-specific features. It maps the features to a multi-dimensional space based on a filter method, and selects the most salient features with the help of Pareto-dominance concepts from multi-objective optimization domain. Our proposed method can be used as online feature selection that deals with problems in which features arrive sequentially while the number of data samples is fixed. In this method, the number of features to be selected is specified during the process of feature selection. However, sometimes it is desired to predefine the number of features. For this reason, an extension of the proposed method is presented to solve this problem. To prove the performance of the proposed methods, several experiments are conducted on some multi-label datasets and the results are compared to five well-established multi-label feature selection methods. The results show the superiority of the proposed methods in terms of different multi-label classification criteria and execution time.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Cheng Liu, Chu-Tao Zheng, Sheng Qian, Si Wu, Hau-San Wong〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Multi-task learning (MTL) aims to enhance generalization performance by exploring the inherent structures across tasks. Most existing MTL methods are based on the assumption that the tasks are positively correlated, and utilize the shared structures among tasks to improve learning performance. By contrast, there also exist competitive structure (negative relationships) among tasks in some real-world applications, and conventional MTL methods which explore shared structures across tasks may lead to unsatisfactory performance in this setting. Another challenge, especially in a high dimensional setting, is to exclude irrelevant features (sparse structure) from the final model. For this purpose, this work propose a new method, which is referred to as Sparse Exclusive Lasso (SpEL) for multi-task learning. The proposed SpEL is able to capture the competitive relationship among tasks (competitive structure), while remove unimportant features which are common across the tasks from the final model (sparse structure). Experimental studies on synthetic and real data indicate that the proposed method can significantly improve learning performance by identifying sparse and task-competitive structures simultaneously.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Daniel Gribel, Thibaut Vidal〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉 〈p〉Minimum sum-of-squares clustering (MSSC) is a widely used clustering model, of which the popular 〈span〉K-means〈/span〉 algorithm constitutes a local minimizer. It is well known that the solutions of 〈span〉K-means〈/span〉 can be arbitrarily distant from the true MSSC global optimum, and dozens of alternative heuristics have been proposed for this problem. However, no other algorithm has been predominantly adopted in the literature. This may be related to differences of computational effort, or to the assumption that a near-optimal solution of the MSSC has only a marginal impact on clustering validity.〈/p〉 〈p〉In this article, we dispute this belief. We introduce an efficient population-based metaheuristic that uses 〈span〉K-means〈/span〉 as a local search in combination with problem-tailored crossover, mutation, and diversification operators. This algorithm can be interpreted as a multi-start 〈span〉K-means〈/span〉, in which the initial center positions are carefully sampled based on the search history. The approach is scalable and accurate, outperforming all recent state-of-the-art algorithms for MSSC in terms of solution quality, measured by the depth of local minima. This enhanced accuracy leads to clusters which are significantly closer to the ground truth than those of other algorithms, for overlapping Gaussian-mixture datasets with a large number of features. Therefore, improved global optimization methods appear to be essential to better exploit the MSSC model in high dimension.〈/p〉 〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Pingping Zhang, Wei Liu, Hongyu Wang, Yinjie Lei, Huchuan Lu〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Street-level scene segmentation aims to label each pixel of street-view images into specific semantic categories. It has been attracting growing interest due to various real-world applications, especially in the area of autonomous driving. However, this pixel-wise labeling task is very challenging under the complex street-level scenes and large-scale object categories. Motivated by the scene layout of street-view images, in this work we propose a novel Spatial Gated Attention (SGA) module, which automatically highlights the attentive regions for pixel-wise labeling, resulting in effective street-level scene segmentation. The proposed module takes as input the multi-scale feature maps based on a Fully Convolutional Network (FCN) backbone, and produces the corresponding attention mask for each feature map. The learned attention masks can neatly highlight the regions of interest while suppress background clutter. Furthermore, we propose an efficient multi-scale feature interaction mechanism which is able to adaptively aggregate the hierarchical features. Based on the proposed mechanism, the features of different levels are adaptively re-weighted according to the local spatial structure and the surrounding contextual information. Consequently, the proposed modules are able to boost standard FCN architectures and result in an enhanced pixel-wise segmentation for street-level scene images. Extensive experiments on three public available street-level benchmarks demonstrate that the proposed Gated Attention Network (GANet) approach achieves consistently superior performance and outperforms the very recent state-of-the-art methods.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Ya Ju Fan〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉The autoencoder is an artificial neural network that performs nonlinear dimension reduction and learns hidden representations of unlabeled data. With a linear transfer function it is similar to the principal component analysis (PCA). While both methods use weight vectors for linear transformations, the autoencoder does not come with any indication similar to the eigenvalues in PCA that are paired with eigenvectors. We propose a novel autoencoder node saliency method that examines whether the features constructed by autoencoders exhibit properties related to known class labels. The supervised node saliency ranks the nodes based on their capability of performing a learning task. It is coupled with the normalized entropy difference (NED). We establish a property for NED values to verify classifying behaviors among the top ranked nodes. By applying our methods to real datasets, we demonstrate their ability to provide indications on the performing nodes and explain the learned tasks in autoencoders.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): J.I. Rilling, J.J. Acuña, P. Nannipieri, F. Cassan, F. Maruyama, M.A. Jorquera〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Since the 1980s, plant growth‒promoting bacteria (PGPB) have been studied as a sustainable alternative to the use of chemical fertilizers to increase crop yields, and effective PGPB have been isolated from diverse plant (〈em〉e.g.〈/em〉, endosphere and phyllosphere) and soil (〈em〉e.g.〈/em〉, rhizosphere) compartments. Despite the promising plant growth promotion results commonly observed under laboratory and greenhouse conditions, the successful application of PGPB under field conditions has been limited, partly by the lack of knowledge of the ecological/environmental factors affecting the colonization, prevalence and activity of beneficial bacteria on crops. It is generally accepted that the effectiveness of PGPB depends on their ability to colonize a niche and compete with the indigenous plant microbiome under agronomic conditions. However, most studies do not include tracking or monitoring of PGPB in the environment after their application, and the beneficial effects on plants are measured by determining biomass‒ and physiology‒related parameters without confirming bacterial colonization. To date, methods based on reporter genes, immunological reactions and nucleic acids have been applied to track or monitor PGPB in seeds, soils or 〈em〉in planta〈/em〉 after inoculation. In this review, we describe, compare and discuss the methods used for tracking and monitoring PGPB, including challenges and perspectives on some novel methods.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: May 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 89〈/p〉 〈p〉Author(s): Yazhou Yang, Marco Loog〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Standard active learning assumes that human annotations are always obtainable whenever new samples are selected. This, however, is unrealistic in many real-world applications where human experts are not readily available at all times. In this paper, we consider the single shot setting: all the required samples should be chosen in a single shot and no human annotation can be exploited during the selection process. We propose a new method, Active Learning through Random Labeling (ALRL), which substitutes single human annotator for multiple, what we will refer to as, pseudo annotators. These pseudo annotators always provide uniform and random labels whenever new unlabeled samples are queried. This random labeling enables standard active learning algorithms to also exhibit the exploratory behavior needed for single shot active learning. The exploratory behavior is further enhanced by selecting the most representative sample via minimizing nearest neighbor distance between unlabeled samples and queried samples. Experiments on real-world datasets demonstrate that the proposed method outperforms several state-of-the-art approaches.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: May 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 89〈/p〉 〈p〉Author(s): Tao Yao, Gang Wang, Lianshan Yan, Xiangwei Kong, Qingtang Su, Caiming Zhang, Qi Tian〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Hashing based cross-media method has been become an increasingly popular technique in facilitating large-scale multimedia retrieval task, owing to its effectiveness and efficiency. Most existing cross-media hashing methods learn hash functions in a batch based mode. However, in practical applications, data points often emerge in a streaming manner, which makes batch based hashing methods loss their efficiency. In this paper, we propose an Online Latent Semantic Hashing (OLSH) method to address this issue. Only newly arriving multimedia data points are utilized to retrain hash functions efficiently and meanwhile preserve the semantic correlations in old data points. Specifically, for learning discriminative hash codes, discrete labels are mapped to a continuous latent semantic space where the relative semantic distances in data points can be measured more accurately. And then, we propose an online optimization scheme towards the challenging task of learning hash functions efficiently on streaming data points, and the computational complexity and memory cost are much less than the size of training dataset at each round. Extensive experiments across many real-world datasets, 〈em〉e.g.〈/em〉 Wiki, Mir-Flickr25K and NUS-WIDE, show the effectiveness and efficiency of the proposed method.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): Jing Ding, Dong Zhu, Qing-Lin Chen, Fei Zheng, Hong-Tao Wang, Yong-Guan Zhu〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Soil fauna plays crucial roles in litter decomposition and nutrient cycling and their associated microbiota contributes to nutrient absorption, health, metabolism and immunity. However, only a few studies have focused on the associated microbiota of soil fauna, and the effects of fertilization on the bacterial community assembly of soil fauna are poorly understood. Here we collected specimens of the collembolan 〈em〉Folsomia quadrioculata〈/em〉 and the soils they lived in from a long-term fertilization experiment (including urea, sewage sludge and chicken manure). The bacterial communities of the collembolans and soils were investigated using 16S rRNA gene amplicon sequencing. A dominant core microbiota consisting of 244 OTUs were identified in the collembolans, of which 41% were shared with the soil microbiota. Community analysis indicated that the assembly of the collembolan bacterial community was a deterministic process, and a close contact of bacterial community was observed in the collembolan microbiome. The collembolan bacterial communities differed significantly from their surrounding soil samples, and their diversity was lower than that of soil microbial community. Furthermore, soil fertilization and in particular application of inorganic fertilizer altered the bacterial community and metabolic potential of the collembolan associated microbiota. Changes in the soil microbial community moreover played an important role in the shift in bacterial community and metabolic potential of the collembolan-associated microbiota. Our results suggested that long-term fertilization significantly contributed to the assembly of the collembolan bacterial community.〈/p〉〈/div〉 〈/div〉 〈h5〉Graphical abstract〈/h5〉 〈div〉〈p〉〈figure〉〈img src="https://ars.els-cdn.com/content/image/1-s2.0-S0038071718304255-fx1.jpg" width="263" alt="Image 1" title="Image 1"〉〈/figure〉〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Rudolf Haraksim, Javier Galbally, Laurent Beslay〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Nowadays, the majority of fingerprint quality, matching and feature extraction algorithms are developed and trained on fingerprints of adults. Accordingly, the processing of children’s fingerprints presents performance issues derived for the most part from: (1) their smaller size and finer ridge structure; (2) their higher variability over time due to the displacement of minutiae induced by growth. The present article is focused on the second factor. The rapid growth of children fingerprints causes a significant displacement of the minutiae points between samples of the same finger acquired with a few years distance from each other. This displacement results in a decrease of the accuracy of fingerprint recognition systems when the reference and probe sample drift apart in time. This effect is known as biometric ageing. In the present study we propose to address this issue by developing and validating a minutiae-based growth model, derived from a database of over 60,000 children’s fingerprints, acquired in real operational conditions, ranging between 5 and 16 years of age, with a time difference between fingerprint pairs re-enrolments of up to 6 years. We analyze two potential application scenarios for the developed growth model. On one hand, we use the model to grow children’s fingerprints in order to spread out the minutiae points to attain sizes similar to those of a sample captured at a later point in time. On the other hand, we apply the model to rejuvenate fingerprints enrolled at a later stage by contracting the minutiae points so that their location is more similar to those of a sample acquired earlier. In both scenarios, the application of the growth model to produce artificially grown/rejuvenated fingerprint minutiae templates results in a significant improvement of the matching scores compared to the ones produced by original fingerprints.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Zhihui Li, Lina Yao, Xiaojun Chang, Kun Zhan, Jiande Sun, Huaxiang Zhang〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Zero-shot complex event detection has been an emerging task in coping with the scarcity of labeled training videos in practice. Aiming to progress beyond the state-of-the-art zero-shot event detection, we propose a new zero-shot event detection approach, which exploits the semantic correlation between an event and concepts. Based on the concept detectors pre-trained from external sources, our method learns the semantic correlation from the concept vocabulary and emphasizes on the most related concepts for the zero-shot event detection. Particularly, a novel Event-Adaptive Concept Integration algorithm is introduced to estimate the effectiveness of semantically related concepts by assigning different weights to them. As opposed to assigning weights by an invariable strategy, we compute the weights of concepts using the area under score curve. The assigned weights are incorporated into the confidence score vector statistically to better characterize the event-concept correlation. Our algorithm is proved to be able to harness the related concepts discriminatively tailored for a target event. Extensive experiments are conducted on the challenging TRECVID event video datasets, which demonstrate the advantage of our approach over the state-of-the-art methods.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Jun Xu, Wangpeng An, Lei Zhang, David Zhang〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉The use of sparse representation (SR) and collaborative representation (CR) for pattern classification has been widely studied in tasks such as face recognition and object categorization. Despite the success of SR/CR based classifiers, it is still arguable whether it is the ℓ〈sub〉1〈/sub〉-norm sparsity or the ℓ〈sub〉2〈/sub〉-norm collaborative property that brings the success of SR/CR based classification. In this paper, we investigate the use of nonnegative representation (NR) for pattern classification, which is largely ignored by previous work. Our analyses reveal that NR can boost the representation power of homogeneous samples while limiting the representation power of heterogeneous samples, making the representation sparse and discriminative simultaneously and thus providing a more effective solution to representation based classification than SR/CR. Our experiments demonstrate that the proposed NR based classifier (NRC) outperforms previous representation based classifiers. With deep features as inputs, it also achieves state-of-the-art performance on various visual classification tasks.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Xin Liu, Jiajia Geng, Haibin Ling, Yiu-ming Cheung〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Speaker naming has recently received considerable attention in identifying the active speaking character in a movie video, and face cue alone is generally insufficient to achieve reliable performance due to its significant appearance variations. In this paper, we treat the speaker naming task as a group of matched audio-face pair finding problems, and present an efficient attention guided deep audio-face fusion approach to detect the active speakers. First, we start with VGG-encoding of face images and extract the Mel-Frequency Cepstrum Coefficients from audio signals. Then, two efficient audio encoding modules, namely two-layer Long Short-Term Memory encoding and two-dimensional convolution encoding, are addressed to discriminate the high-level audio features. Meanwhile, we train an end-to-end audio-face common attention model to discriminate the face attention vector, featuring adaptively to accommodate various face variations. Further, an efficient factorized bilinear model is presented to deeply fuse the paired audio-face features, whereby the joint audio-face representation can be reliably obtained for speaker naming. Extensive experiments highlight the superiority of the proposed approach and show its very competitive performance with the state-of-the-arts.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Raymond Ptucha, Felipe Petroski Such, Suhas Pillai, Frank Brockler, Vatsala Singh, Paul Hutkowski〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉The recognition of handwritten text is challenging as there are virtually infinite ways a human can write the same message. Deep learning approaches for handwriting analysis have recently demonstrated breakthrough performance using both lexicon-based architectures and recurrent neural networks. This paper presents a fully convolutional network architecture which outputs arbitrary length symbol streams from handwritten text. A preprocessing step normalizes input blocks to a canonical representation which negates the need for costly recurrent symbol alignment correction. When a lexicon is known, we further introduce a probabilistic character error rate to correct errant word blocks. Our multi-state convolutional method is the first to demonstrate state-of-the-art results on both lexicon-based and arbitrary symbol based handwriting recognition benchmarks.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Yong Shi, Minglong Lei, Hong Yang, Lingfeng Niu〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉In network embedding, random walks play a fundamental role in preserving network structures. However, random walk methods have two limitations. First, they are unstable when either the sampling frequency or the number of node sequences changes. Second, in highly biased networks, random walks are likely to bias to high-degree nodes and neglect the global structure information. To solve the limitations, we present in this paper a network diffusion embedding method. To solve the first limitation, our method uses a diffusion driven process to capture both depth and breadth information in networks. Temporal information is also included into node sequences to strengthen information preserving. To solve the second limitation, our method uses the network inference method based on information diffusion cascades to capture the global network information. Experiments show that the new proposed method is more robust to highly unbalanced networks and well performed when sampling under each node is rare.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): Michael Philben, Jianqiu Zheng, Markus Bill, Jeffrey M. Heikoop, George Perkins, Ziming Yang, Stan D. Wullschleger, David E. Graham, Baohua Gu〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Increasing nitrogen (N) availability in Arctic soils could stimulate the growth of both plants and microorganisms by relieving the constraints of nutrient limitation. It was hypothesized that organic N addition to anoxic tundra soil would increase CH〈sub〉4〈/sub〉 production by stimulating the fermentation of labile substrates, which is considered the rate-limiting step in anaerobic C mineralization. We tested this hypothesis through both field and lab-based experiments. In the field experiment, we injected a solution of 〈sup〉13〈/sup〉C- and 〈sup〉15〈/sup〉N-labeled glutamate 35 cm belowground at a site near Nome on the Seward Peninsula, Alaska, and observed the resulting changes in porewater geochemistry and dissolved greenhouse gas concentrations. The concentration of free glutamate declined rapidly within hours of injection, and the 〈sup〉15〈/sup〉N label was recovered almost exclusively as dissolved organic N within 62 h. These results indicate rapid microbial assimilation of the added N and transformation into novel organic compounds. We observed increasing concentrations of dissolved CH〈sub〉4〈/sub〉 and Fe(II), indicating rapid stimulation of methanogenesis and Fe(III) reduction. Low molecular weight organic acids such as acetate and propionate accumulated despite increasing consumption through anaerobic C mineralization. A laboratory soil column flow experiment using active layer soil collected from the same site further supported these findings. Glutamate recovery was low compared to a conservative bromide tracer, but concentrations of NO〈sub〉3〈/sub〉〈sup〉−〈/sup〉 and NH〈sub〉4〈/sub〉〈sup〉+〈/sup〉 remained low, consistent with microbial uptake of the added N. Similar to the field experiment, we observed both increasing Fe(II) and organic acid concentrations. Together, these results support our hypothesis of increased fermentation in response to organic N addition and suggest that increasing N availability could accelerate CH〈sub〉4〈/sub〉 production in tundra soils.〈/p〉〈/div〉 〈/div〉 〈h5〉Graphical abstract〈/h5〉 〈div〉〈p〉〈figure〉〈img src="https://ars.els-cdn.com/content/image/1-s2.0-S003807171830419X-fx1.jpg" width="215" alt="Image 1" title="Image 1"〉〈/figure〉〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Tomislav Pribanić, Tomislav Petković, Matea Đonlić〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉3D registration is a very active topic, spanning research areas such as computational geometry, computer graphics and pattern recognition. It aims to solve spatial transformation that aligns two point clouds. In this work we propose the use of a single direction sensor, such as an accelerometer or a magnetometer, commonly available on contemporary mobile platforms, such as tablets and smartphones. Both sensors have been heavily investigated earlier, but only for joint use with other sensors, such as gyroscopes and GPS. We show a time-efficient and accurate 3D registration method that takes advantage of only either an accelerometer or a magnetometer. We demonstrate a 3D reconstruction of individual point clouds and the proposed 3D registration method on a tablet equipped with an accelerometer or a magnetometer. However, we point out that the proposed method is not restricted to mobile platforms. Indeed, it can easily be applied in any 3D measurement system that is upgradable with some ubiquitous direction sensor, for example by adding a smartphone equipped with either an accelerometer or a magnetometer. We compare the proposed method against several state-of-the-art methods implemented in the open source Point Cloud Library (PCL). The proposed method outperforms the PCL methods tested, both in terms of processing time and accuracy.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): Srabani Das, Brian K. Richards, Kelly L. Hanley, Leilah Krounbi, M.F. Walter, M. Todd Walter, Tammo S. Steenhuis, Johannes Lehmann〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉The effect of long-term versus short-term water content on soil organic carbon (SOC) mineralizability was evaluated in a six-week incubation trial. Soils were sampled from field sites in upstate New York used for rain-fed bioenergy crop production: nitrogen (N)- fertilized reed canarygrass, switchgrass, switchgrass + N, as well as a broadleaf-grass fallow. Within each cropping system, natural moisture gradients due to topography and subsoil structure allowed us to sample across regions with high (0.5 g g〈sup〉−1〈/sup〉), mid (0.4 g g〈sup〉−1〈/sup〉) and low (0.3 g g〈sup〉−1〈/sup〉) water content. Moisture of the laboratory incubations was adjusted mimicking the three average field moisture levels in a full factorial design. Increasing laboratory moisture in the incubations increased cumulative carbon mineralization per unit soil (C mineralization) and cumulative C mineralization per unit SOC (C mineralizability) (main effect p 〈 0.0001), indicating that lower average moisture as found at this site on average limited mineralization but higher average moisture did not. C mineralizability at high field moisture was 31% (25-42%) lower than at low field moisture across all cropping systems, regardless of moisture adjustment in the incubation. The mean slow C pool size of soils from high field moisture sites (997.1 ± 0.1 mg C g〈sup〉−1〈/sup〉 C) was 0.2% greater than that of soils from low field moisture sites (p 〈 0.0001), obtained by fitting a double-exponential model. The mean residence time of the slow mineralizing C pool for soils from low field moisture sites was 5.5 ± 0.1 years, in comparison to 8.0 ± 0.1 years for soils from high field moisture sites (p 〈 0.0001). While permanganate-oxidizable carbon (POXC) per unit SOC (r = 0.1) was positively correlated to C mineralizability, wet aggregate stability (r = −0.2) was negatively correlated to C mineralizability. Above-ground biomass did not affect C mineralizability (p 〉 0.05) and root biomass marginally influenced (p = 0.05) C mineralizability after correcting for soil texture variations. Additionally, after correcting for soil texture variations and biomass inputs, C mineralizability significantly decreased with higher field moisture (p = 0.02), indicating possible stabilization mechanisms through mineral interactions of SOC under high water content. Bulk contents of pedogenic iron and aluminum determined by oxalate extraction did not clearly explain differences in mineralizability. However, exchangeable calcium and magnesium contents were significantly (p 〈 0.0001) greater in high moisture soils than soils with lower moisture. Additionally, cumulative C mineralizability at 6 weeks was negatively correlated to calcium (r = −0.7) and magnesium (r = −0.6) and mean residence time of the modeled slow pool correlated positively with calcium (r = 0.4). Therefore, cation bridging by retained or illuviated base ions was more important than redox changes of iron as a stabilization mechanism in this experiment.〈/p〉〈/div〉 〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Danyu Lai, Wei Tian, Long Chen〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉In this paper, we propose a novel and efficient multi-stage approach, which combines both semi-supervised learning and fine-grained learning to improve the performance of classification model learned only from a few samples. The fine-grained category recognition process utilized in our method is dubbed as MSR. In this process, we cut images into multi-scaled parts to feed into the network to learn more fine-grained features. By assigning these image cuts with dynamic weights, we can reduce the negative impact of background information and thus achieve a more accurate prediction. Furthermore, we present the voted pseudo label (VPL) which is an efficient method of semi-supervised learning. In this approach, for unlabeled data, VPL picks up the classes with non-confused labels verified by the consensus prediction of different classification models. These two methods can be applied to most neural network models and training methods. Inspired from classifier-based adaptation, we also propose a mix deep CNN architecture (MixDCNN). Both the VPL and MSR are integrated with the MixDCNN. Comprehensive experiments demonstrate the effectiveness of VPL and MSR. Without bottles and jars, we achieve the state-of-the-art or even better performance in two fine-grained recognition tasks on the datasets of Stanford Dogs and CUB Birds, with the accuracy of 95.6% and 85.2%, respectively.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: April 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Pattern Recognition, Volume 88〈/p〉 〈p〉Author(s): Qiang Wang, Huijie Fan, Gan Sun, Yang Cong, Yandong Tang〈/p〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Recently, generative adversarial networks (GANs) have demonstrated high-quality reconstruction in face completion. There is still much room for improvement over the conventional GAN models that do not explicitly address the texture details problem. In this paper, we propose a Laplacian-pyramid-based generative framework for face completion. This framework can produce more realistic results (1) by deriving precise content information of missing face regions in a coarse-to-fine fashion and (2) by propagating the high-frequency details from the surrounding area via a modified residual learning model. Specifically, for the missing regions, we design a Laplacian-pyramid-based convolutional network framework that can predict missing regions under different resolutions; this framework takes advantage of multiscale features shared from low levels and extracted from middle layers for the next finer level. For high-frequency details, we construct a new residual learning network to eliminate color discrepancies between the missing and surrounding regions progressively. Furthermore, a multiloss function is proposed to supervise the generative process. To optimize the model, we train the entire generative model with deep supervision using a joint reconstruction loss, which ensures that the generated image is as realistic as the original. Extensive experiments on benchmark datasets show that the proposed framework exhibits superior performance over state-of-the-art methods in terms of predictive accuracy, both quantitatively and qualitatively.〈/p〉〈/div〉

Beschreibung: 〈p〉Publication date: March 2019〈/p〉 〈p〉〈b〉Source:〈/b〉 Soil Biology and Biochemistry, Volume 130〈/p〉 〈p〉Author(s): Yuxi Guo, Meixiang Gao, Jie Liu, Andrey S. Zaitsev, Donghui Wu〈/p〉 〈div xml:lang="en"〉 〈h5〉Abstract〈/h5〉 〈div〉〈p〉Research on factors determining soil metacommunity patterns across multiple spatial scales is quite rare. In this study, we aimed to compare the mechanisms that drive species co-occurrence and consequently form the ground-dwelling macro-arthropod metacommunity structure at local (10〈sup〉4〈/sup〉 m) and regional (10〈sup〉6〈/sup〉 m) scales. For the comparison, we used three distant (approximately 200 km from each other) locations within the northeast black soil region of China. At each location, we had five sampling plots (at a distance of 500 m from each other), and at each plot, we had five sampling sites within agrolandscapes (10 m away from each other). At each site, we set three pitfall traps. Animals were collected three times: in May, July and September 2015. The analysis of the elements of metacommunity structure showed that the regional level of the metacommunity always demonstrated a Clementsian structure (a grouped distribution of species along environmental gradients), while the local scale metacommunity structure was dependent on sampling month and varied between Clementsian, random and nested distributions. Based on the results of a variance partitioning analysis, including pure environmental (environmental filtering) and pure spatial predictors (dispersal), as well as the interaction of environmental and spatial factors (shared fraction), we observed that the shared fraction at the regional scale was 30% larger than at the local scale. A species co-occurrence analysis further demonstrated that the observed C-scores were not significantly higher than simulated indices for each guild separately at the local scale. Our results suggest that biotic interactions are not the key drivers of metacommunity structure at the local scale in agriculture, while environmental filtering and dispersal appear to be key drivers of metacommunity structure at the regional scale.〈/p〉〈/div〉 〈/div〉