Anaerobic oxidation of ethane by archaea from a marine hydrocarbon seep.

Ethane is the second most abundant component of natural gas in addition to methane, and-similar to methane-is chemically unreactive. The biological consumption of ethane under anoxic conditions was suggested by geochemical profiles at marine hydrocarbon seeps1-3, and through ethane-dependent sulfate reduction in slurries4-7. Nevertheless, the microorganisms and reactions that catalyse this process have to date remained unknown8. Here we describe ethane-oxidizing archaea that were obtained by specific enrichment over ten years, and analyse these archaea using phylogeny-based fluorescence analyses, proteogenomics and metabolite studies. The co-culture, which oxidized ethane completely while reducing sulfate to sulfide, was dominated by an archaeon that we name 'Candidatus Argoarchaeum ethanivorans'; other members were sulfate-reducing Deltaproteobacteria. The genome of Ca. Argoarchaeum contains all of the genes that are necessary for a functional methyl-coenzyme M reductase, and all subunits were detected in protein extracts. Accordingly, ethyl-coenzyme M (ethyl-CoM) was identified as an intermediate by liquid chromatography-tandem mass spectrometry. This indicated that Ca. Argoarchaeum initiates ethane oxidation by ethyl-CoM formation, analogous to the recently described butane activation by 'Candidatus Syntrophoarchaeum'9. Proteogenomics further suggests that oxidation of intermediary acetyl-CoA to CO2 occurs through the oxidative Wood-Ljungdahl pathway. The identification of an archaeon that uses ethane (C2H6) fills a gap in our knowledge of microorganisms that specifically oxidize members of the homologous alkane series (CnH2n+2) without oxygen. Detection of phylogenetic and functional gene markers related to those of Ca. Argoarchaeum at deep-sea gas seeps10-12 suggests that archaea that are able to oxidize ethane through ethyl-CoM are widespread members of the local communities fostered by venting gaseous alkanes around these seeps.

Microbial diversity and oil biodegradation potential of northern Barents Sea sediments.

The Arctic, an essential ecosystem on Earth, is subject to pronounced anthropogenic pressures, most notable being the climate change and risks of crude oil pollution. As crucial elements of Arctic environments, benthic microbiomes are involved in climate-relevant biogeochemical cycles and hold the potential to remediate upcoming contamination. Yet, the Arctic benthic microbiomes are among the least explored biomes on the planet. Here we combined geochemical analyses, incubation experiments, and microbial community profiling to detail the biogeography and biodegradation potential of Arctic sedimentary microbiomes in the northern Barents Sea. The results revealed a predominance of bacterial and archaea phyla typically found in the deep marine biosphere, such as Chloroflexi, Atribacteria, and Bathyarcheaota. The topmost benthic communities were spatially structured by sedimentary organic carbon, lacking a clear distinction among geographic regions. With increasing sediment depth, the community structure exhibited stratigraphic variability that could be correlated to redox geochemistry of sediments. The benthic microbiomes harbored multiple taxa capable of oxidizing hydrocarbons using aerobic and anaerobic pathways. Incubation of surface sediments with crude oil led to proliferation of several genera from the so-called rare biosphere. These include Alkalimarinus and Halioglobus, previously unrecognized as hydrocarbon-degrading genera, both harboring the full genetic potential for aerobic alkane oxidation. These findings increase our understanding of the taxonomic inventory and functional potential of unstudied benthic microbiomes in the Arctic.

The Great Oxidation Event expanded the genetic repertoire of arsenic metabolism and cycling.

The rise of oxygen on the early Earth about 2.4 billion years ago reorganized the redox cycle of harmful metal(loids), including that of arsenic, which doubtlessly imposed substantial barriers to the physiology and diversification of life. Evaluating the adaptive biological responses to these environmental challenges is inherently difficult because of the paucity of fossil records. Here we applied molecular clock analyses to 13 gene families participating in principal pathways of arsenic resistance and cycling, to explore the nature of early arsenic biogeocycles and decipher feedbacks associated with planetary oxygenation. Our results reveal the advent of nascent arsenic resistance systems under the anoxic environment predating the Great Oxidation Event (GOE), with the primary function of detoxifying reduced arsenic compounds that were abundant in Archean environments. To cope with the increased toxicity of oxidized arsenic species that occurred as oxygen built up in Earth's atmosphere, we found that parts of preexisting detoxification systems for trivalent arsenicals were merged with newly emerged pathways that originated via convergent evolution. Further expansion of arsenic resistance systems was made feasible by incorporation of oxygen-dependent enzymatic pathways into the detoxification network. These genetic innovations, together with adaptive responses to other redox-sensitive metals, provided organisms with novel mechanisms for adaption to changes in global biogeocycles that emerged as a consequence of the GOE.

Genome and proteome analyses show the gaseous alkane degrader Desulfosarcina sp. strain BuS5 as an extreme metabolic specialist.

The metabolic potential of the sulfate-reducing bacterium Desulfosarcina sp. strain BuS5, currently the only pure culture able to oxidize the volatile alkanes propane and butane without oxygen, was investigated via genomics, proteomics and physiology assays. Complete genome sequencing revealed that strain BuS5 encodes a single alkyl-succinate synthase, an enzyme which apparently initiates oxidation of both propane and butane. The formed alkyl-succinates are oxidized to CO2 via beta oxidation and the oxidative Wood-Ljungdahl pathways as shown by proteogenomics analyses. Strain BuS5 conserves energy via the canonical sulfate reduction pathway and electron bifurcation. An ability to utilize long-chain fatty acids, mannose and oligopeptides, suggested by automated annotation pipelines, was not supported by physiology assays and in-depth analyses of the corresponding genetic systems. Consistently, comparative genomics revealed a streamlined BuS5 genome with a remarkable paucity of catabolic modules. These results establish strain BuS5 as an exceptional metabolic specialist, able to grow only with propane and butane, for which we propose the name Desulfosarcina aeriophaga BuS5. This highly restrictive lifestyle, most likely the result of habitat-driven evolutionary gene loss, may provide D. aeriophaga BuS5 a competitive edge in sediments impacted by natural gas seeps. Etymology: Desulfosarcina aeriophaga, aério (Greek): gas; phágos (Greek): eater; D. aeriophaga: a gas eating or gas feeding Desulfosarcina.

Intramolecular isotopic evidence for bacterial oxidation of propane in subsurface natural gas reservoirs.

Microbial anaerobic oxidation of hydrocarbons is a key process potentially involved in a myriad of geological and biochemical environments yet has remained notoriously difficult to identify and quantify in natural environments. We performed position-specific carbon isotope analysis of propane from cracking and incubation experiments. Anaerobic bacterial oxidation of propane leads to a pronounced and previously unidentified 13C enrichment in the central position of propane, which contrasts with the isotope signature associated with the thermogenic process. This distinctive signature allows the detection and quantification of anaerobic oxidation of hydrocarbons in diverse natural gas reservoirs and suggests that this process may be more widespread than previously thought. Position-specific isotope analysis can elucidate the fate of natural gas hydrocarbons and provide insight into a major but previously cryptic process controlling the biogeochemical cycling of globally significant greenhouse gases.

High Arsenic Levels Increase Activity Rather than Diversity or Abundance of Arsenic Metabolism Genes in Paddy Soils.

Arsenic (As) metabolism genes are generally present in soils, but their diversity, relative abundance, and transcriptional activity in response to different As concentrations remain unclear, limiting our understanding of the microbial activities that control the fate of an important environmental pollutant. To address this issue, we applied metagenomics and metatranscriptomics to paddy soils showing a gradient of As concentrations to investigate As resistance genes (ars) including arsR, acr3, arsB, arsC, arsM, arsI, arsP, and arsH as well as energy-generating As respiratory oxidation (aioA) and reduction (arrA) genes. Somewhat unexpectedly, the relative DNA abundances and diversities of ars, aioA, and arrA genes were not significantly different between low and high (∼10 versus ∼100 mg kg-1) As soils. Compared to available metagenomes from other soils, geographic distance rather than As levels drove the different compositions of microbial communities. Arsenic significantly increased ars gene abundance only when its concentration was higher than 410 mg kg-1. In contrast, metatranscriptomics revealed that relative to low-As soils, high-As soils showed a significant increase in transcription of ars and aioA genes, which are induced by arsenite, the dominant As species in paddy soils, but not arrA genes, which are induced by arsenate. These patterns appeared to be community wide as opposed to taxon specific. Collectively, our findings advance understanding of how microbes respond to high As levels and the diversity of As metabolism genes in paddy soils and indicated that future studies of As metabolism in soil or other environments should include the function (transcriptome) level. IMPORTANCE Arsenic (As) is a toxic metalloid pervasively present in the environment. Microorganisms have evolved the capacity to metabolize As, and As metabolism genes are ubiquitously present in the environment even in the absence of high concentrations of As. However, these previous studies were carried out at the DNA level; thus, the activity of the As metabolism genes detected remains essentially speculative. Here, we show that the high As levels in paddy soils increased the transcriptional activity rather than the relative DNA abundance and diversity of As metabolism genes. These findings advance our understanding of how microbes respond to and cope with high As levels and have implications for better monitoring and managing an important toxic metalloid in agricultural soils and possibly other ecosystems.

Organic Carbon Amendments Affect the Chemodiversity of Soil Dissolved Organic Matter and Its Associations with Soil Microbial Communities.

The "4 per mil" initiative recognizes the pivotal role of soil in carbon resequestration. The need for evidence to substantiate the influence of agricultural practices on chemical nature of soil carbon and microbial biodiversity has become a priority. However, owing to the molecular complexity of soil dissolved organic matter (DOM), specific linkages to microbial biodiversity have eluded researchers. Here, we characterized the chemodiversity of soil DOM, assessed the variation of soil bacterial community composition (BCC), and identified specific linkages between DOM traits and BCC. Sustained organic carbon amendment significantly ( P < 0.05) increased total organic matter reservoirs, resulted in higher chemodiversity of DOM and emergence of recalcitrant moieties (H/C < 1.5). In the meantime, sustained organic carbon amendment shaped the BCC to a more eutrophic state while long-term chemical fertilization directed the BCC toward an oligotrophic state. Meanwhile, higher connectivity and complexity were observed in organic carbon amendment by DOM-BCC network analysis, indicating that soil microbes tended to have more interaction with DOM molecules after organic matter inputs. These results highlight the potential for organic carbon amendments to not only build soil carbon stocks and increase their resilience but also mediate the functional state of soil bacterial communities.

Molecular Chemodiversity of Dissolved Organic Matter in Paddy Soils.

Organic matter (OM), and dissolved organic matter (DOM), have a major influence upon biogeochemical processes; most significantly, the carbon cycle. To date, very few studies have examined the spatial heterogeneity of DOM in paddy soils. Thus, very little is known about the DOM molecular profiles and the key environmental factors that underpin DOM molecular chemodiversity in paddy soils. Here, Fourier-Transform Ion Cyclotron Resonance Mass Spectrometry was applied to unambiguously resolve 11 361 molecular formulas in 16 paddy soils; thereby elucidating the molecular characteristics of paddy soil DOM. Soil pH, iron complexing index (Fep/FeR) and C/N ratio were established to be key factors controlling DOM profiles. Polycyclic aromatics (derived from combustion) and polyphenols (derived from plants) increased with increasing pH, while polyphenols molecules, pyrogenic aromatics, and carboxylic compounds decreased with increasing iron complexing index. Patterns in molecular profiles indicated DOM in paddy soils to become more recalcitrant at higher soil C/N ratio and higher pH. Furthermore, plant-derived polyphenols and pyrogenic DOM were retained favorably by iron and the chemodiversity of DOM in paddy soil increased with increasing soil C/N ratios. This study provides critical information about DOM characteristics at a molecular level and will inform better global management of soil carbon in paddy soil ecosystems.

Heavy Metal Induced Antibiotic Resistance in Bacterium LSJC7.

Co-contamination of antibiotics and heavy metals prevails in the environment, and may play an important role in disseminating bacterial antibiotic resistance, but the selective effects of heavy metals on bacterial antibiotic resistance is largely unclear. To investigate this, the effects of heavy metals on antibiotic resistance were studied in a genome-sequenced bacterium, LSJC7. The results showed that the presence of arsenate, copper, and zinc were implicated in fortifying the resistance of LSJC7 towards tetracycline. The concentrations of heavy metals required to induce antibiotic resistance, i.e., the minimum heavy metal concentrations (MHCs), were far below (up to 64-fold) the minimum inhibition concentrations (MIC) of LSJC7. This finding indicates that the relatively low heavy metal levels in polluted environments and in treated humans and animals might be sufficient to induce bacterial antibiotic resistance. In addition, heavy metal induced antibiotic resistance was also observed for a combination of arsenate and chloramphenicol in LSJC7, and copper/zinc and tetracycline in antibiotic susceptible strain Escherichia coli DH5α. Overall, this study implies that heavy metal induced antibiotic resistance might be ubiquitous among various microbial species and suggests that it might play a role in the emergence and spread of antibiotic resistance in metal and antibiotic co-contaminated environments.

Artificial light at night causes community instability of bacterial community in urban soils.

Urban soils host diverse bacteria crucial for ecosystem functions and urban health. As urbanization rises, artificial light at night (ALAN) imposes disturbances on soil ecosystems, yet how ALAN affects the structure and stability of soil bacterial community remains unclear. Here we coupled a short-term incubation experiment, community profiling, network analysis, and in situ field survey to assess the ecological impacts of ALAN. We showed that ALAN influenced bacterial compositions and shifted the bacterial network to a less stable phase, altering denitrification potential. Such transition in community stability probably resulted from an ALAN-induced decrease in competition and/or an increase in facilitation, in line with the Stress Gradient Hypothesis. Similar destabilizing effects were also detected in bacterial networks in multiple urban soils subjected to different levels of ALAN stress, supporting the action of ALAN on naturally-occurring soil bacterial communities. Overall, our findings highlight ALAN as a new form of anthropogenic stress that jeopardizes the stability of soil bacterial community, which would facilitate ecological projection of expanding ALAN exposure.

Contrastive Prototype-Guided Generation for Generalized Zero-Shot Learning.

Generalized zero-shot learning (GZSL) aims to recognize both seen and unseen classes, while only samples from seen classes are available for training. The mainstream methods mitigate the lack of unseen training data by simulating the visual unseen samples. However, the sample generator is actually learned with just seen-class samples, and semantic descriptions of unseen classes are just provided to the pre-trained sample generator for unseen data generation, therefore, the generator would have bias towards seen categories, and the unseen generation quality, including both precision and diversity, is still the main learning challenge. To this end, we propose a Prototype-Guided Generation for Generalized Zero-Shot Learning (PGZSL), in order to guide the sample generation with unseen knowledge. First, unseen data generation is guided and rectified in PGZSL by contrastive prototypical anchors with both class semantic consistency and feature discriminability. Second, PGZSL introduces Certainty-Driven Mixup for generator to enrich the diversity of generated unseen samples, while suppress the generation of uncertain boundary samples as well. Empirical results over five benchmark datasets show that PGZSL significantly outperforms the SOTA methods in both ZSL and GZSL tasks.

Archaea oxidizing alkanes through alkyl-coenzyme M reductases.

This review synthesizes recent discoveries of novel archaea clades capable of oxidizing higher alkanes, from volatile ones like ethane to longer-chain alkanes like hexadecane. These archaea, termed anaerobic multicarbon alkane-oxidizing archaea (ANKA), initiate alkane oxidation using alkyl-coenzyme M reductases, enzymes similar to the methyl-coenzyme M reductases of methanogenic and anaerobic methanotrophic archaea (ANME). The polyphyletic alkane-oxidizing archaea group (ALOX), encompassing ANME and ANKA, harbors increasingly complex alkane degradation pathways, correlated with the alkane chain length. We discuss the evolutionary trajectory of these pathways emphasizing metabolic innovations and the acquisition of metabolic modules via lateral gene transfer. Additionally, we explore the mechanisms by which archaea couple alkane oxidation with the reduction of electron acceptors, including electron transfer to partner sulfate-reducing bacteria (SRB). The phylogenetic and functional constraints that shape ALOX-SRB associations are also discussed. We conclude by highlighting the research needs in this emerging research field and its potential applications in biotechnology.

Nitrogen fertilization modulates rice phyllosphere functional genes and pathogens through fungal communities.

The phyllosphere is a vital yet often neglected habitat hosting diverse microorganisms with various functions. However, studies regarding how the composition and functions of the phyllosphere microbiome respond to agricultural practices, like nitrogen fertilization, are limited. This study investigated the effects of long-term nitrogen fertilization with different levels (CK, N90, N210, N330) on the functional genes and pathogens of the rice phyllosphere microbiome. Results showed that the relative abundance of many microbial functional genes in the rice phyllosphere was significantly affected by nitrogen fertilization, especially those involved in C fixation and denitrification genes. Different nitrogen fertilization levels have greater effects on fungal communities than bacteria communities in the rice phyllosphere, and network analysis and structural equation models further elucidate that fungal communities not only changed bacterial-fungal inter-kingdom interactions in the phyllosphere but also contributed to the variation of biogeochemical cycle potential. Besides, the moderate nitrogen fertilization level (N210) was associated with an enrichment of beneficial microbes in the phyllosphere, while also resulting in the lowest abundance of pathogenic fungi (1.14 %). In contrast, the highest abundance of pathogenic fungi (1.64 %) was observed in the highest nitrogen fertilization level (N330). This enrichment of pathogen due to high nitrogen level was also regulated by the fungal communities, as revealed through SEM analysis. Together, we demonstrated that the phyllosphere fungal communities were more sensitive to the nitrogen fertilization levels and played a crucial role in influencing phyllosphere functional profiles including element cycling potential and pathogen abundance. This study expands our knowledge regarding the role of phyllosphere fungal communities in modulating the element cycling and plant health in sustainable agriculture.

Convex Subspace Clustering by Adaptive Block Diagonal Representation.

Subspace clustering is a class of extensively studied clustering methods where the spectral-type approaches are its important subclass. Its key first step is to desire learning a representation coefficient matrix with block diagonal structure. To realize this step, many methods were successively proposed by imposing different structure priors on the coefficient matrix. These impositions can be roughly divided into two categories, i.e., indirect and direct. The former introduces the priors such as sparsity and low rankness to indirectly or implicitly learn the block diagonal structure. However, the desired block diagonalty cannot necessarily be guaranteed for noisy data. While the latter directly or explicitly imposes the block diagonal structure prior such as block diagonal representation (BDR) to ensure so-desired block diagonalty even if the data is noisy but at the expense of losing the convexity that the former's objective possesses. For compensating their respective shortcomings, in this article, we follow the direct line to propose adaptive BDR (ABDR) which explicitly pursues block diagonalty without sacrificing the convexity of the indirect one. Specifically, inspired by Convex BiClustering, ABDR coercively fuses both columns and rows of the coefficient matrix via a specially designed convex regularizer, thus naturally enjoying their merits and adaptively obtaining the number of blocks. Finally, experimental results on synthetic and real benchmarks demonstrate the superiority of ABDR to the state-of-the-arts (SOTAs).

Multiplane Convex Proximal Support Vector Machine.

As an effective method for XOR problems, generalized eigenvalue proximal support vector machine (GEPSVM) recently has gained widespread attention accompanied with many variants proposed. Although these variants strengthen the classification performance to different extents, the number of fitting hyperplanes, similar to GEPSVM, for each class is still limited to just one. Intuitively, using single hyperplane seems not enough, especially for the datasets with complex feature structures. Therefore, this article mainly focuses on extending the fitting hyperplanes for each class from single one to multiple ones. However, such an extension from the original GEPSVM is not trivial even though, if possible, the elegant solution via generalized eigenvalues will also not be guaranteed. To address this issue, we first make a simple yet crucial transformation for the optimization problem of GEPSVM and then propose a novel multiplane convex proximal support vector machine (MCPSVM), where a set of hyperplanes determined by the features of the data are learned for each class. We adopt a strictly (geodesically) convex objective to characterize this optimization problem; thus, a more elegant closed-form solution is obtained, which only needs a few lines of MATLAB codes. Besides, MCPSVM is more flexible in form and can be naturally and seamlessly extended to the feature weighting learning, whereas GEPSVM and its variants can hardly straightforwardly work like this. Extensive experiments on benchmark and large-scale image datasets indicate the advantages of our MCPSVM.

Partial Domain Adaptation Without Domain Alignment.

Unsupervised domain adaptation (UDA) aims to transfer knowledge from a well-labeled source domain to a related and unlabeled target domain with identical label space. The main workhorse in UDA is domain alignment and has proven successful. However, it is practically difficult to find an appropriate source domain with identical label space. A more practical scenario is partial domain adaptation (PDA) where the source label space subsumes the target one. Unfortunately, due to the non-identity between label spaces, it is extremely hard to obtain an ideal alignment, conversely, easier resulting in mode collapse and negative transfer. These motivate us to find a relatively simpler alternative to solve PDA. To achieve this, we first explore a theoretical analysis, which says that the target risk is bounded by both model smoothness and between-domain discrepancy. Then, we instantiate the model smoothness as an intra-domain structure preserving (IDSP) while giving up possibly riskier domain alignment. To our best knowledge, this is the first naive attempt for PDA without alignment. Finally, our empirical results on benchmarks demonstrate that IDSP is not only superior to the PDA SOTAs (e.g.,  âˆ¼ +10% on Cl → Rw and  âˆ¼ +8% on Ar → Rw), but also complementary to domain alignment in the standard UDA.

Universum-Inspired Supervised Contrastive Learning.

As an effective data augmentation method, Mixup synthesizes an extra amount of samples through linear interpolations. Despite its theoretical dependency on data properties, Mixup reportedly performs well as a regularizer and calibrator contributing reliable robustness and generalization to deep model training. In this paper, inspired by Universum Learning which uses out-of-class samples to assist the target tasks, we investigate Mixup from a largely under-explored perspective - the potential to generate in-domain samples that belong to none of the target classes, that is, universum. We find that in the framework of supervised contrastive learning, Mixup-induced universum can serve as surprisingly high-quality hard negatives, greatly relieving the need for large batch sizes in contrastive learning. With these findings, we propose Universum-inspired supervised Contrastive learning (UniCon), which incorporates Mixup strategy to generate Mixup-induced universum as universum negatives and pushes them apart from anchor samples of the target classes. We extend our method to the unsupervised setting, proposing Unsupervised Universum-inspired contrastive model (Un-Uni). Our approach not only improves Mixup with hard labels, but also innovates a novel measure to generate universum data. With a linear classifier on the learned representations, UniCon shows state-of-the-art performance on various datasets. Specially, UniCon achieves 81.7% top-1 accuracy on CIFAR-100, surpassing the state of art by a significant margin of 5.2% with a much smaller batch size, typically, 256 in UniCon vs. 1024 in SupCon (Khosla et al., 2020) using ResNet-50. Un-Uni also outperforms SOTA methods on CIFAR-100. The code of this paper is released on https://github.com/hannaiiyanggit/UniCon.

Dual-Correction-Adaptation Network for Noisy Knowledge Transfer.

Unsupervised domain adaptation (UDA) promotes target learning via a single-directional transfer from label-rich source domain to unlabeled target, while its reverse adaption from target to source has not been jointly considered yet. In real teaching practice, a teacher helps students learn and also gets promotion from students, and such a virtuous cycle inspires us to explore dual-directional transfer between domains. In fact, target pseudo-labels predicted by source commonly involve noise due to model bias; moreover, source domain usually contains innate noise, which inevitably aggravates target noise, leading to noise amplification. Transfer from target to source exploits target knowledge to rectify the adaptation, consequently enables better source transfer, and exploits a virtuous transfer circle. To this end, we propose a dual-correction-adaptation network (DualCAN), in which adaptation and correction cycle between domains, such that learning in both domains can be boosted gradually. To the best of our knowledge, this is the first naive attempt of dual-directional adaptation. Empirical results validate DualCAN with remarkable performance gains, particularly for extreme noisy tasks (e.g., approximately + 10 % on D → A of Office-31 with 40 % label corruption).

TIToK: A solution for bi-imbalanced unsupervised domain adaptation.

Unsupervised domain adaptation (UDA) aims to transfer knowledge via domain alignment, and typically assumes balanced data distribution. When deployed in real tasks, however, (i) each domain usually suffers from class imbalance, and (ii) different domains may have different class imbalance ratios. In such bi-imbalanced cases with both within-domain and across-domain imbalance, source knowledge transfer may degenerate the target performance. Some recent efforts have adopted source re-weighting to this issue, in order to align label distributions across domains. However, since target label distribution is unknown, the alignment might be incorrect or even risky. In this paper, we propose an alternative solution named TIToK for bi-imbalanced UDA, by directly Transferring Imbalance-Tolerant Knowledge across domains. In TIToK, a class contrastive loss is presented for classification, in order to alleviate the sensitivity to imbalance in knowledge transfer. Meanwhile, knowledge of class correlation is transferred as a supplementary, which is commonly invariant to imbalance. Finally, discriminative feature alignment is developed for a more robust classifier boundary. Experiments over benchmark datasets show that TIToK achieves competitive performance with the state-of-the-arts, and its performance is less sensitive to imbalance.

Human Collective Intelligence Inspired Multi-View Representation Learning - Enabling View Communication by Simulating Human Communication Mechanism.

In real-world applications, we often encounter multi-view learning tasks where we need to learn from multiple sources of data or use multiple sources of data to make decisions. Multi-view representation learning, which can learn a unified representation from multiple data sources, is a key pre-task of multi-view learning and plays a significant role in real-world applications. Accordingly, how to improve the performance of multi-view representation learning is an important issue. In this work, inspired by human collective intelligence shown in group decision making, we introduce the concept of view communication into multi-view representation learning. Furthermore, by simulating human communication mechanism, we propose a novel multi-view representation learning approach that can fulfill multi-round view communication. Thus, each view of our approach can exploit the complementary information from other views to help with modeling its own representation, and mutual help between views is achieved. Extensive experiment results on six datasets from three significant fields indicate that our approach substantially improves the average classification accuracy by 4.536% in medicine and bioinformatics fields as well as 4.115% in machine learning field.