ABSTRACT
Stipa breviflora is a dominant species in the desert steppe of Northern China. Grazing is the main land use pattern of grassland, which could cause a variety of adaptive evolutionary mechanisms in plant community composition as well as individual plant growth and morphological characteristics. However, very little is known about the morphological structure and transcriptional regulation response to different grazing intensities in S. breviflora. In this study, transcriptome and anatomical analyses of S. breviflora under different grazing intensities, including no grazing, moderate grazing, and heavy grazing, were performed. The anatomical analysis results showed that epidermis cells and xylems significantly thicken with grazing intensity, suggesting that grazing results in increasing lignification. Furthermore, the components of cell walls such as lignin, cellulose, hemicellulose, and pectin were all increased dramatically and significantly under both moderate and heavy grazing. Transcriptome analysis showed that the differentially expressed genes related to different grazing intensities were also engaged in plant cell wall formation and in photosynthesis and respiration. In addition, the activities of ATP synthase and Rubisco-activating enzyme increased significantly with enhanced grazing intensity and differed significantly between moderate and heavy grazing intensities. The trends in transcriptome and plant phenotype changes are consistent. Taken together, these results indicated that S. breviflora has evolved a grazing tolerance strategy under long-term grazing conditions, influencing photosynthesis and respiration in terms of its own structure and enzyme activities in the body, to maintain normal life activities under different grazing conditions.
ABSTRACT
Smart windows always respond to single stimuli, which cannot satisfy various needs in practical applications. Smart windows that integrate thermotropic, electrochromic and power-generating functions in one device is highly challenging yet important in satisfying on-demand light modulation and energy efficiency in practical applications. Herein, a thermoresponsive lower critical solution temperature (LCST) ion gel was fabricated via a facile in situ polymerization of butyl acrylate in a conventional ionic liquid to explore "all in one" smart windows. The ion gel-assembled smart windows are thermotropic and electrochromic with a reliable adjustment of light transparency as well as power-generating, enabled by the ionic Soret effect of ionic liquids. Additionally, the ion gels demonstrated self-defensive robust mechanical properties, thermal insulating and antifogging properties. With such an interdisciplinary and comprehensive study of the ion gels, the LCST ion gels could fulfil the requirements of genius windows with high energy-saving potential and exceptional climate adaptability, such as shut-down of light transmission in summer, daily solar energy collection, and colour changes on demand. It conceptually updates smart windows from an energy saving to an energy supplier in buildings. It is the first time to explore the "all in one" smart windows based on integrated multifunctional ionic liquids, which could greatly bridge the gap between the materials and buildings to accelerate practical applications of smart windows.
ABSTRACT
Inhalable chemicals found in children's play mats can be slowly released into indoor environments and consequently threaten human health. In this study, the partition coefficients of seven inhalable chemicals between play mats and air were calculated by headspace gas chromatography-mass spectrometry based on the law of conservation of mass and the principle of equilibrium of headspace bottles. Furthermore, an emission source model for the residual ratio of the inhalable chemicals in play mats was established. Most substances found in play mats have large partition coefficients owing to the complex void structure of the mats, which adsorbs a large number of organic pollutants. The partition coefficient is not only related to the boiling point and environmental temperature, but also the specific material and the adsorption of the organic pollutant onto the material. The emission source model for children's play mats developed in this study can characterize the decay of the inhalable chemicals over time. The data showed that after eight days of placing the play mat in a ventilated environment, the residual ratio of seven inhalable chemicals did not exceed 15 %.
Subject(s)
Environmental Pollutants , Child , Humans , AdsorptionABSTRACT
OBJECTIVE: This study aims to extend the Theory of Planned Behavior with Triggers (TPBT) to improve the prediction of physical activity (PA) behavior using the TPB model. METHODS: Questionnaires, including the TPB scale, PA rating scale (PARS-3), and triggers scale, were administered to 596 Chinese college students, and the data were analyzed using SPSS 23.0 and AMOS 24.0. RESULTS: Subjective norm (SN), attitude (AT), and perceived behavioral control (PBC) all significantly and positively affected behavioral intention (BI). The path coefficient of PA behavior was significantly influenced by the interaction term of three types of triggers and BI, and the TPB with Triggers (TPBT) model improved the explanation rate of PA behavior. CONCLUSION: Triggers have a moderating effect on the relationship between BI and PA behavior, and the TPBT model better explains college students' PA behavior. Among the three dimensions of triggers, people are more receptive to facilitator and signal triggers than spark triggers. This has practical implications for practitioners designing interventions to promote PA among college students.
ABSTRACT
Nitrogen and phosphorus resorption (NRE and PRE) is a critical nutrient conservation mechanism maintaining plant growth in already disturbed barren ecosystems. The complexity of plant nutrient resorption variations in long-term grazing regions is regulated by plant traits, nutritional utilization strategies, and soil conditions following changes in grazing patterns. Therefore, a detailed investigation into their underlying mechanism is still required. Here we investigated leaf nutrient concentration and resorption in dominant species Cleistogenes songorica (C. squarrosa) and Stipa breviflora (S. breviflora) response to 15-years continuous grazing (moderate and heavy grazing) in desert steppe. Moderate grazing enhanced green leaf N and P content in C. songorica and partially increased N content in S. breviflora. Heavy grazing consistently increased N content in C. songorica, but its P content as well as N and P content in S. breviflora were largely stable. Moderate grazing enhanced NRE but unaffected PRE in both S. breviflora and C. songorica. Heavy grazing reduced NRE and PRE in C. songorica. Although soil variables (nutrients and moisture) did not affect foliar nutrients, it's a key driver of nutrient resorption efficiency. Of all measured influence factors, soil moisture is the one most important and negatively correlated with NRE and PRE in S. breviflora. While it was not observed in C. songorica. In S. breviflora, its NRE was adversely linked with soil N, in addition, both NRE and PRE were positively associated with green leaf nutrients. Senesced leaf nutrients are the predominant factor influencing nutrient resorption efficiency in C. songorica, which were adversely associated. Overall, our results indicate significant variations in nutrient resorption efficiency patterns between the two dominant species due to divergent plant adaptation strategies to grazing and the local environment. The foliar nutritional status and soil conditions may play significant roles in regulating nutrient resorption in arid long-term grazing desert steppe.
Subject(s)
Ecosystem , Nitrogen Isotopes , Soil , Poaceae/physiology , Plants , Nitrogen/analysis , Nutrients , Phosphorus , Plant Leaves/chemistryABSTRACT
World models learn the consequences of actions in vision-based interactive systems. However, in practical scenarios like autonomous driving, noncontrollable dynamics that are independent or sparsely dependent on action signals often exist, making it challenging to learn effective world models. To address this issue, we propose Iso-Dream++, a model-based reinforcement learning approach that has two main contributions. First, we optimize the inverse dynamics to encourage the world model to isolate controllable state transitions from the mixed spatiotemporal variations of the environment. Second, we perform policy optimization based on the decoupled latent imaginations, where we roll out noncontrollable states into the future and adaptively associate them with the current controllable state. This enables long-horizon visuomotor control tasks to benefit from isolating mixed dynamics sources in the wild, such as self-driving cars that can anticipate the movement of other vehicles, thereby avoiding potential risks. On top of our previous work (Pan et al. 2022), we further consider the sparse dependencies between controllable and noncontrollable states, address the training collapse problem of state decoupling, and validate our approach in transfer learning setups. Our empirical study demonstrates that Iso-Dream++ outperforms existing reinforcement learning models significantly on CARLA and DeepMind Control.
ABSTRACT
The diffusion characteristics of fine-grained gangue slurry are significant for gangue consumption and ecological environment protection, especially in underground mines. Therefore, based on the theory of porous medium transportation, this paper establishes a 3D caving zone model, proposes methods for reconstructing, statistical, and analytical voids, characterizes the distribution of voids in the caving zone and the diffusion characteristics of gangue slurry, and explores the influencing factors of the distribution of voids and the diffusion characteristics of gangue slurry. The diffusion rate and depth of the gangue slurry were calculated using Df and DT parameters. Finally, a fluidization filling method for gangue was proposed, providing a theoretical basis for solid waste treatment and environmental protection in coal mines.
Subject(s)
Coal , Mining , CavesABSTRACT
Learning predictive models for unlabeled spatiotemporal data is challenging in part because visual dynamics can be highly entangled, especially in real scenes. In this paper, we refer to the multi-modal output distribution of predictive learning as spatiotemporal modes. We find an experimental phenomenon named spatiotemporal mode collapse (STMC) on most existing video prediction models, that is, features collapse into invalid representation subspaces due to the ambiguous understanding of mixed physical processes. We propose to quantify STMC and explore its solution for the first time in the context of unsupervised predictive learning. To this end, we present ModeRNN, a decoupling-aggregation framework that has a strong inductive bias of discovering the compositional structures of spatiotemporal modes between recurrent states. We first leverage a set of dynamic slots with independent parameters to extract individual building components of spatiotemporal modes. We then perform a weighted fusion of slot features to adaptively aggregate them into a unified hidden representation for recurrent updates. Through a series of experiments, we show high correlation between STMC and the fuzzy prediction results of future video frames. Besides, ModeRNN is shown to better mitigate STMC and achieve the state of the art on five video prediction datasets.
ABSTRACT
Invited for the cover of this issue are Chunpu Li, Hong Liu and co-workers at Shanghai Institute of Materia Medica, Nanjing University of Chinese Medicine, and Hangzhou Institute for Advanced Study. The image depicts rhodium catalysis converting the readily available podophyllotoxin into four kinds of novel derivatives. Read the full text of the article at 10.1002/chem.202300960.
ABSTRACT
A divergent synthesis of podophyllotoxin derivatives from simple and readily available starting materials through a late-stage functionalization strategy by rhodium catalysis is reported here. This strategy uses the ketone and oxime in substrates as directing groups. Four kinds of novel podophyllotoxin derivatives have been obtained without any erosion of the enantiopurity, thus indicating the broad substrate scope of this method. Additionally, by using the newly developed strategy, 9 aa, which exhibited excellent anticancer activity, can be prepared by a sequential transformation. In particularly, 9 aa suppressed HeLa cells with IC50 values of 74.5â nM, thus providing a promising lead compound for future drug discovery.
ABSTRACT
We consider a new problem of adapting a human mesh reconstruction model to out-of-domain streaming videos, where the performance of existing SMPL-based models is significantly affected by the distribution shift represented by different camera parameters, bone lengths, backgrounds, and occlusions. We tackle this problem through online adaptation, gradually correcting the model bias during testing. There are two main challenges: First, the lack of 3D annotations increases the training difficulty and results in 3D ambiguities. Second, non-stationary data distribution makes it difficult to strike a balance between fitting regular frames and hard samples with severe occlusions or dramatic changes. To this end, we propose the Dynamic Bilevel Online Adaptation algorithm (DynaBOA). It first introduces the temporal constraints to compensate for the unavailable 3D annotations and leverages a bilevel optimization procedure to address the conflicts between multi-objectives. DynaBOA provides additional 3D guidance by co-training with similar source examples retrieved efficiently despite the distribution shift. Furthermore, it can adaptively adjust the number of optimization steps on individual frames to fully fit hard samples and avoid overfitting regular frames. DynaBOA achieves state-of-the-art results on three out-of-domain human mesh reconstruction benchmarks.
ABSTRACT
The predictive learning of spatiotemporal sequences aims to generate future images by learning from the historical context, where the visual dynamics are believed to have modular structures that can be learned with compositional subsystems. This paper models these structures by presenting PredRNN, a new recurrent network, in which a pair of memory cells are explicitly decoupled, operate in nearly independent transition manners, and finally form unified representations of the complex environment. Concretely, besides the original memory cell of LSTM, this network is featured by a zigzag memory flow that propagates in both bottom-up and top-down directions across all layers, enabling the learned visual dynamics at different levels of RNNs to communicate. It also leverages a memory decoupling loss to keep the memory cells from learning redundant features. We further propose a new curriculum learning strategy to force PredRNN to learn long-term dynamics from context frames, which can be generalized to most sequence-to-sequence models. We provide detailed ablation studies to verify the effectiveness of each component. Our approach is shown to obtain highly competitive results on five datasets for both action-free and action-conditioned predictive learning scenarios.
ABSTRACT
A mechanically induced artificial potential barrier (MIAPB) in piezoelectric semiconductor devices is set up under the action of a pair of tensile/compressive mechanical loadings. Three factors, namely, the barrier height, width and position, affect the nature and extent of interaction between the MIAPB and the contact barrier, and the tuning characteristics (generated under conditions of the artificial barrier) of the piezoelectric PN junctions were studied. The influence of these factors resulted in variations in the interaction intensities, superposition effects, carrier inversion degrees and carrier redistribution ranges. Subsequently, the limit tuning effects exerted by the tensile/compressive-mode MIAPB on the PN junctions were studied. The inconsistency between the left and right end of the tensile-mode MIAPB under conditions of the offset loading state proves that the maximum tuning effect is generated when both sides of the interface are symmetrically loaded. The range of carrier redistribution and the over-inversion of local carriers, affected by the width and height of MIAPB, result in a second competitive mechanism. The carrier redistribution range and the carrier inversion degree require that the compressive-mode MIAPB be sufficiently wide. The interaction intensities and the superposition effects, affected by the position and height of the MIAPB, contribute to the second competing mechanism. We logically clarify the relationship between multiple competition and find that the emergence of multiple competitive mechanisms proves the existence of the limit tuning effect of MIAPB on the I-V properties of PN junctions. The results reported herein provide a platform for understanding the mechanical tuning laws governing the functions of piezoelectric PN junctions and piezoelectric devices.
ABSTRACT
Multi-view 3D reconstruction generally adopts the feature fusion strategy to guide the generation of 3D shape for objects with different views. Empirically, the correspondence learning of object regions across different views enables better feature fusion. However, such idea has not been fully exploited in existing methods. Furthermore, current methods fail to explore the intrinsic dependency among regions within a 3D shape, leading to a rough reconstruction result. To address the above issues, we propose a Dual-View 3D Point Cloud reconstruction architecture named DVPC, which takes two views images as inputs, and progressively generates a refined 3D point cloud. First, a point cloud generation network is assigned to generate a coarse point cloud for each input view. Second, a dual-view point clouds synthesis network is presented in DVPC. It constructs a regional attention mechanism to learn a high-quality correspondence among regions across two coarse point clouds in different views, so that our DVPC can achieve feature fusion accurately. And then it develops a point cloud deformation module to produce a relatively-precise point cloud via establishing the communication between the coarse point cloud and the fused feature. Lastly, a point-region transformer network is devised to model the dependency among regions within the relatively-precise point cloud. With the dependency, the relatively-precise point cloud is refined into a desirable 3D point cloud with rich details. Qualitative and quantitative experiments on the ShapeNet and Pix3D datasets demonstrate that the proposed DVPC outperforms the state-of-the-art methods in terms of reconstruction quality.
ABSTRACT
This study discusses the interaction between electromechanical fields and carriers in a multilayered ZnO beam where the c-axis of every two adjacent layers is alternately opposite along the thickness direction. A multi-field coupling model is proposed from the Timoshenko beam theory together with the phenomenological theory of piezoelectric semiconductors, including Gauss's law and the continuity equation of currents. The analytical solutions are obtained for a bent beam with different numbers of layers. Numerical results show that polarized charges occur at the interfaces between every two adjacent layers due to the opposite electromechanical coupling effects. It was found that a series of alternating potential-barrier/well structures are induced by the polarized charges, which can be used to forbid the passing of low-energy mobile charges. Moreover, it was also observed that the induced polarized charges could weaken the shielding effect of carrier redistribution. These results are useful for the design of piezotronic devices.
ABSTRACT
In Tangshan Mine, there are four engineering problems, including the layout of multiple seam mining system, serious environmental damages caused by gangue accumulation and surface collapse, nearly 100 million tons of three under coal (coal trapped under buildings, water bodies, and railways) to be released, and the control accuracy of stratum formations to be improved. The work developed the integrated green mining technology of coal mining, gangue washing-backfilling, strata control, and system monitoring. First, a production system was designed for surface-underground transportation of backfilling materials, underground gangue separation, multiple-seam combined backfilling, and mining. Second, key backfilling equipment was developed at mining heights of 2.2 and 3.5 m by backfilling system layout and precise stratum control methods in Tangshan Mine. After real-time monitoring of stope pressure, backfilling effect, and surface deformation, we evaluated the implementation effect of mining, washing, and backfilling technology. The integrated coal mining in seams 5, 8, and 9 of Tangshan Mine showed that the four backfilling surfaces of T3281, T3292, F5001, and F5002 recovered 946,000 t of raw coal and 1.18 million tons of filled gangue, with a net profit of 363.20 million yuan. The maximum land subsidence values (18, 119, 64, and 47mm) were far lower than the deformation extremes based on the requirements of surface building protection.
Subject(s)
Coal Mining , China , Coal , TechnologyABSTRACT
BACKGROUND: This study aimed to develop a Physical Activity Triggers Questionnaire for Chinese college students and to evaluate its reliability and validity. METHODS: On the theoretical basis of the Fogg behavior model and semi-open interviews, an initial questionnaire with 18 items was compiled. The initial questionnaire was administered to 575 students, and to examine its reliability and validity, item discrimination analysis, correlation analysis, homogeneity test, and exploratory factor analysis were conducted using SPSS 26.0. After the examination of the initial questionnaire, the initial 18 items were reduced to 14. The 14-item questionnaire was administered to 621 college students, and with the data, correlation analysis, confirmatory factor analysis, validity test, and reliability test were conducted. RESULTS: To examine the psychometric properties of the 18 items, exploratory factor analysis and confirmatory factor analysis were conducted, and their reliability and validity were examined. After the first round of item development analysis, four items were removed, and a triggers questionnaire with 14 items was developed. The 14 items had three dimensions, including spark, signal, and facilitator triggers, and the cumulative explained variance of the three dimensions was 61.21%. The confirmatory factor analysis of the three dimensions of the 14 items indicated appropriate scale fit indices. The internal consistency reliability, split-half reliability, and test-retest reliability of the 14 items were 0.925, 0.821, and 0.860, respectively, showing that the items have appropriate reliability. CONCLUSIONS: The Physical Activity Triggers Questionnaire of the study has acceptable reliability and validity. It is the first questionnaire to measure Chinese college students' triggers of physical activity and will provide a new basis for the understanding of psychometric properties of physical activity triggers. In addition, the future findings collected from the developed triggers questionnaire can be used to develop strategies to promote health among college students.
ABSTRACT
Unsupervised domain adaptation (UDA) aims to transfer knowledge from a related but different well-labeled source domain to a new unlabeled target domain. Most existing UDA methods require access to the source data, and thus are not applicable when the data are confidential and not shareable due to privacy concerns. This paper aims to tackle a realistic setting with only a classification model available trained over, instead of accessing to, the source data. To effectively utilize the source model for adaptation, we propose a novel approach called Source HypOthesis Transfer (SHOT), which learns the feature extraction module for the target domain by fitting the target data features to the frozen source classification module (representing classification hypothesis). Specifically, SHOT exploits both information maximization and self-supervised learning for the feature extraction module learning to ensure the target features are implicitly aligned with the features of unseen source data via the same hypothesis. Furthermore, we propose a new labeling transfer strategy, which separates the target data into two splits based on the confidence of predictions (labeling information), and then employ semi-supervised learning to improve the accuracy of less-confident predictions in the target domain. We denote labeling transfer as SHOT++ if the predictions are obtained by SHOT. Extensive experiments on both digit classification and object recognition tasks show that SHOT and SHOT++ achieve results surpassing or comparable to the state-of-the-arts, demonstrating the effectiveness of our approaches for various visual domain adaptation problems. Code will be available at https://github.com/tim-learn/SHOT-plus.
Subject(s)
Algorithms , Neural Networks, ComputerABSTRACT
This paper introduces video domain generalization where most video classification networks degenerate due to the lack of exposure to the target domains of divergent distributions. We observe that the global temporal features are less generalizable, due to the temporal domain shift that videos from other unseen domains may have an unexpected absence or misalignment of the temporal relations. This finding has motivated us to solve video domain generalization by effectively learning the local-relation features of different timescales that are more generalizable, and exploiting them along with the global-relation features to maintain the discriminability. This paper presents the VideoDG framework with two technical contributions. The first is a new deep architecture named the Adversarial Pyramid Network, which improves the generalizability of video features by capturing the local-relation, global-relation, and cross-relation features progressively. On the basis of pyramid features, the second contribution is a new and robust approach of adversarial data augmentation that can bridge different video domains by improving the diversity and quality of augmented data. We construct three video domain generalization benchmarks in which domains are divided according to different datasets, different consequences of actions, or different camera views, respectively. VideoDG consistently outperforms the combinations of previous video classification models and existing domain generalization methods on all benchmarks.
ABSTRACT
Genotyping platforms, as critical supports for genomics, genetics, and molecular breeding, have been well implemented at national institutions/universities in developed countries and multinational seed companies that possess high-throughput, automatic, large-scale, and shared facilities. In this study, we integrated an improved genotyping by target sequencing (GBTS) system with capture-in-solution (liquid chip) technology to develop a multiple single-nucleotide polymorphism (mSNP) approach in which mSNPs can be captured from a single amplicon. From one 40K maize mSNP panel, we developed three types of markers (40K mSNPs, 251K SNPs, and 690K haplotypes), and generated multiple panels with various marker densities (1K-40K mSNPs) by sequencing at different depths. Comparative genetic diversity analysis was performed with genic versus intergenic markers and di-allelic SNPs versus non-typical SNPs. Compared with the one-amplicon-one-SNP system, mSNPs and within-mSNP haplotypes are more powerful for genetic diversity detection, linkage disequilibrium decay analysis, and genome-wide association studies. The technologies, protocols, and application scenarios developed for maize in this study will serve as a model for the development of mSNP arrays and highly efficient GBTS systems in animals, plants, and microorganisms.