Different chemical compositions of soil organic carbon(SOC)affect its persistence and whether it signifi-cantly differs between natural forests and plantations remains unclear.By synthesizing 234 observations of SOC c...Different chemical compositions of soil organic carbon(SOC)affect its persistence and whether it signifi-cantly differs between natural forests and plantations remains unclear.By synthesizing 234 observations of SOC chemical compositions,we evaluated global patterns of concentra-tion,individual chemical composition(alkyl C,O-alkyl C,aromatic C,and carbonyl C),and their distribution even-ness.Our results indicate a notably higher SOC,a markedly larger proportion of recalcitrant alkyl C,and lower easily decomposed carbonyl C proportion in natural forests.How-ever,SOC chemical compositions were appreciably more evenly distributed in plantations.Based on the assumed con-ceptual index of SOC chemical composition evenness,we deduced that,compared to natural forests,plantations may have higher possible resistance to SOC decomposition under disturbances.In tropical regions,SOC levels,recalcitrant SOC chemical composition,and their distributed evenness were significantly higher in natural forests,indicating that SOC has higher chemical stability and possible resistance to decomposition.Climate factors had minor effects on alkyl C in forests globally,while they notably affected SOC chemi-cal composition in tropical forests.This could contribute to the differences in chemical compositions and their distrib-uted evenness between plantations and natural stands.展开更多
Among hyperspectral imaging technologies, interferometric spectral imaging is widely used in remote sening due to advantages of large luminous flux and high resolution. However, with complicated mechanism, interferome...Among hyperspectral imaging technologies, interferometric spectral imaging is widely used in remote sening due to advantages of large luminous flux and high resolution. However, with complicated mechanism, interferometric imaging faces the impact of multi-stage degradation. Most exsiting interferometric spectrum reconstruction methods are based on tradition model-based framework with multiple steps, showing poor efficiency and restricted performance. Thus, we propose an interferometric spectrum reconstruction method based on degradation synthesis and deep learning.Firstly, based on imaging mechanism, we proposed an mathematical model of interferometric imaging to analyse the degradation components as noises and trends during imaging. The model consists of three stages, namely instrument degradation, sensing degradation, and signal-independent degradation process. Then, we designed calibration-based method to estimate parameters in the model, of which the results are used for synthesizing realistic dataset for learning-based algorithms.In addition, we proposed a dual-stage interferogram spectrum reconstruction framework, which supports pre-training and integration of denoising DNNs. Experiments exhibits the reliability of our degradation model and synthesized data, and the effectiveness of the proposed reconstruction method.展开更多
This article presents hourly load forecasting by using an incremental learning model called Online Sequential Extreme Learning Machine(OSELM),which can learn and adapt automatically according to new arrival input.Howe...This article presents hourly load forecasting by using an incremental learning model called Online Sequential Extreme Learning Machine(OSELM),which can learn and adapt automatically according to new arrival input.However,the use of OS-ELM requires a sufficient amount of initial training sample data,which makes OS-ELM inoperable if sufficiently accurate sample data cannot be obtained.To solve this problem,a synthesis of the initial training sample is proposed.The synthesis of the initial sample is achieved by taking the first data received at the start of working and adding random noises to that data to create new and sufficient samples.Then the synthesis samples are used to initial train the OS-ELM.This proposed method is compared with Fully Online Extreme Learning Machine(FOS-ELM),which is an incremental learning model that also does not require the initial training samples.Both the proposed method and FOS-ELM are used for hourly load forecasting from the Hourly Energy Consumption dataset.Experiments have shown that the proposed method with a wide range of noise levels,can forecast hourly load more accurately than the FOS-ELM.展开更多
While recent years have witnessed a dramatic upsurge of exploiting deep neural networks toward solving image denoising,existing methods mostly rely on simple noise assumptions,such as additive white Gaussian noise(AWG...While recent years have witnessed a dramatic upsurge of exploiting deep neural networks toward solving image denoising,existing methods mostly rely on simple noise assumptions,such as additive white Gaussian noise(AWGN),JPEG compression noise and camera sensor noise,and a general-purpose blind denoising method for real images remains unsolved.In this paper,we attempt to solve this problem from the perspective of network architecture design and training data synthesis.Specifically,for the network architecture design,we propose a swin-conv block to incorporate the local modeling ability of residual convolutional layer and non-local modeling ability of swin transformer block,and then plug it as the main building block into the widely-used image-to-image translation UNet architecture.For the training data synthesis,we design a practical noise degradation model which takes into consideration different kinds of noise(including Gaussian,Poisson,speckle,JPEG compression,and processed camera sensor noises)and resizing,and also involves a random shuffle strategy and a double degradation strategy.Extensive experiments on AGWN removal and real image denoising demonstrate that the new network architecture design achieves state-of-the-art performance and the new degradation model can help to significantly improve the practicability.We believe our work can provide useful insights into current denoising research.The source code is available at https://github.com/cszn/SCUNet.展开更多
Purpose:We aimed to perform a systematic review and meta-analysis of the effects of training to muscle failure or non-failure on muscular strength and hypertrophy.Methods:Meta-analyses of effect sizes(ESs)explored the...Purpose:We aimed to perform a systematic review and meta-analysis of the effects of training to muscle failure or non-failure on muscular strength and hypertrophy.Methods:Meta-analyses of effect sizes(ESs)explored the effects of training to failure vs.non-failure on strength and hypertrophy.Subgroup meta-analyses explored potential moderating effects of variables such as training status(trained vs.untrained),training volume(volume equated vs.volume non-equated),body region(upper vs.lower),exercise selection(multi-vs.single-joint exercises(only for strength)),and study design(independent vs.dependent groups).Results:Fifteen studies were included in the review.All studies included young adults as participants.Meta-analysis indicated no significant difference between the training conditions for muscular strength(ES=-0.09,95%confidence interval(95%CI):-0.22 to 0.05)and for hypertrophy(ES=0.22,95%CI:-0.11 to 0.55).Subgroup analyses that stratified the studies according to body region,exercise selection,or study design showed no significant differences between training conditions.In studies that did not equate training volume between the groups,the analysis showed significant favoring of non-failure training on strength gains(ES=-0.32,95%CI:-0.57 to-0.07).In the subgroup analysis for resistance-trained individuals,the analysis showed a significant effect of training to failure for muscle hypertrophy(ES=0.15,95%CI:0.03-0.26).Conclusion:Training to muscle failure does not seem to be required for gains in strength and muscle size.However,training in this manner does not seem to have detrimental effects on these adaptations,either.More studies should be conducted among older adults and highly trained individuals to improve the generalizability of these findings.展开更多
Purpose:This review aimed to synthesize previous findings on the test-retest reliability of the 30-15 Intermittent Fitness Test(IFT).Methods:The literature searches were performed in 8 databases.Studies that examined ...Purpose:This review aimed to synthesize previous findings on the test-retest reliability of the 30-15 Intermittent Fitness Test(IFT).Methods:The literature searches were performed in 8 databases.Studies that examined the test-retest reliability of the 30-15 IFT and presented the intraclass correlation coefficient(ICC) and/or the coefficient of variation(CV) for maximal velocity and/or peak heart rate were included.The consensus-based standards for the selection of health measurement instruments(COSMIN) checklist was used for the assessment of the methodological quality of the included studies.Results:Seven studies,with a total of 10 study groups,explored reliability of maximal velocity assessed by the 30-15 IFT.ICCs ranged from0.80 to 0.99,where 70% of ICCs were≥0.90.CVs for maximal velocity ranged from 1.5% to 6.0%.Six studies,with a total of 7 study groups,explored reliability of peak heart rate as assessed by the 30-15 IFT.ICCs ranged from 0.90 to 0.97(i.e.,all ICCs were≥0.90).CVs ranged from 0.6% to 4.8%.All included studies were of excellent methodological quality.Conclusion:From the results of this systematic review,it can be concluded that the 30-15 IFT has excellent test-retest reliability for both maximal velocity and peak heart rate.The test may,therefore,be used as a reliable measure of fitness in research and sports practice.展开更多
Due to the lack of parallel data in current grammatical error correction(GEC)task,models based on sequence to sequence framework cannot be adequately trained to obtain higher performance.We propose two data synthesis ...Due to the lack of parallel data in current grammatical error correction(GEC)task,models based on sequence to sequence framework cannot be adequately trained to obtain higher performance.We propose two data synthesis methods which can control the error rate and the ratio of error types on synthetic data.The first approach is to corrupt each word in the monolingual corpus with a fixed probability,including replacement,insertion and deletion.Another approach is to train error generation models and further filtering the decoding results of the models.The experiments on different synthetic data show that the error rate is 40%and that the ratio of error types is the same can improve the model performance better.Finally,we synthesize about 100 million data and achieve comparable performance as the state of the art,which uses twice as much data as we use.展开更多
The standard approach to tackling computer vision problems is to train deep convolutional neural network(CNN)models using large-scale image datasets that are representative of the target task.However,in many scenarios...The standard approach to tackling computer vision problems is to train deep convolutional neural network(CNN)models using large-scale image datasets that are representative of the target task.However,in many scenarios,it is often challenging to obtain sufficient image data for the target task.Data augmentation is a way to mitigate this challenge.A common practice is to explicitly transform existing images in desired ways to create the required volume and variability of training data necessary to achieve good generalization performance.In situations where data for the target domain are not accessible,a viable workaround is to synthesize training data from scratch,i.e.,synthetic data augmentation.This paper presents an extensive review of synthetic data augmentation techniques.It covers data synthesis approaches based on realistic 3D graphics modelling,neural style transfer(NST),differential neural rendering,and generative modelling using generative adversarial networks(GANs)and variational autoencoders(VAEs).For each of these classes of methods,we focus on the important data generation and augmentation techniques,general scope of application and specific use-cases,as well as existing limitations and possible workarounds.Additionally,we provide a summary of common synthetic datasets for training computer vision models,highlighting the main features,application domains and supported tasks.Finally,we discuss the effectiveness of synthetic data augmentation methods.Since this is the first paper to explore synthetic data augmentation methods in great detail,we are hoping to equip readers with the necessary background information and in-depth knowledge of existing methods and their attendant issues.展开更多
Prosodic control is an important part of speech synthesis system. Prosodic parameters choice right or wrong influences the quality of synthetic speech directly. At present, text to speech system has less effective des...Prosodic control is an important part of speech synthesis system. Prosodic parameters choice right or wrong influences the quality of synthetic speech directly. At present, text to speech system has less effective describe to reflect data relationships in the corpus. A new research approach - data mining technology to discover those relationships by association rules modeling is presented. And a new algorithm for generating association rules of prosodic parameters including pitch parameters and duration parameters from corpus is developed. The output rules improve the correctness of syllable choice in text to speech system.展开更多
Classification of sheep behaviour from a sequence of tri-axial accelerometer data has the potential to enhance sheep management.Sheep behaviour is inherently imbalanced(e.g.,more ruminating than walking)resulting in u...Classification of sheep behaviour from a sequence of tri-axial accelerometer data has the potential to enhance sheep management.Sheep behaviour is inherently imbalanced(e.g.,more ruminating than walking)resulting in underperforming classification for the minority activities which hold importance.Existing works have not addressed class imbalance and use traditional machine learning techniques,e.g.,Random Forest(RF).We investigated Deep Learning(DL)models,namely,Long Short Term Memory(LSTM)and Bidirectional LSTM(BLSTM),appropriate for sequential data,from imbalanced data.Two data sets were collected in normal grazing conditions using jaw-mounted and earmounted sensors.Novel to this study,alongside typical single classes,e.g.,walking,depending on the behaviours,data samples were labelled with compound classes,e.g.,walking_-grazing.The number of steps a sheep performed in the observed 10 s time window was also recorded and incorporated in the models.We designed several multi-class classification studies with imbalance being addressed using synthetic data.DL models achieved superior performance to traditional ML models,especially with augmented data(e.g.,4-Class+Steps:LSTM 88.0%,RF 82.5%).DL methods showed superior generalisability on unseen sheep(i.e.,F1-score:BLSTM 0.84,LSTM 0.83,RF 0.65).LSTM,BLSTM and RF achieved sub-millisecond average inference time,making them suitable for real-time applications.The results demonstrate the effectiveness of DL models for sheep behaviour classification in grazing conditions.The results also demonstrate the DL techniques can generalise across different sheep.The study presents a strong foundation of the development of such models for real-time animal monitoring.展开更多
Area and test time are two major overheads encountered duringdata path high level synthesis for BIST. This paper presents an approach to behavioral synthesis for loop-based BIST. By taking into account the requirement...Area and test time are two major overheads encountered duringdata path high level synthesis for BIST. This paper presents an approach to behavioral synthesis for loop-based BIST. By taking into account the requirements of theBIST scheme during behavioral synthesis processes, an area optimal BIST solutioncan be obtained. This approach is based on the use of test resources reusabilitythat results in a fewer number of registers being modified to be test registers. Thisis achieved by incorporating self-testability constraints during register assignmentoperations. Experimental results on benchmarks are presented to demonstrate theeffectiveness of the approach.展开更多
基金supported by the National Natural Science Foundation of China(Grants 31971463,31930078)the National Key R&D Program of China(Grant 2021YFD2200402)the Chinese Academy of Forestry(Grant CAFYBB2020ZA001).
文摘Different chemical compositions of soil organic carbon(SOC)affect its persistence and whether it signifi-cantly differs between natural forests and plantations remains unclear.By synthesizing 234 observations of SOC chemical compositions,we evaluated global patterns of concentra-tion,individual chemical composition(alkyl C,O-alkyl C,aromatic C,and carbonyl C),and their distribution even-ness.Our results indicate a notably higher SOC,a markedly larger proportion of recalcitrant alkyl C,and lower easily decomposed carbonyl C proportion in natural forests.How-ever,SOC chemical compositions were appreciably more evenly distributed in plantations.Based on the assumed con-ceptual index of SOC chemical composition evenness,we deduced that,compared to natural forests,plantations may have higher possible resistance to SOC decomposition under disturbances.In tropical regions,SOC levels,recalcitrant SOC chemical composition,and their distributed evenness were significantly higher in natural forests,indicating that SOC has higher chemical stability and possible resistance to decomposition.Climate factors had minor effects on alkyl C in forests globally,while they notably affected SOC chemi-cal composition in tropical forests.This could contribute to the differences in chemical compositions and their distrib-uted evenness between plantations and natural stands.
文摘Among hyperspectral imaging technologies, interferometric spectral imaging is widely used in remote sening due to advantages of large luminous flux and high resolution. However, with complicated mechanism, interferometric imaging faces the impact of multi-stage degradation. Most exsiting interferometric spectrum reconstruction methods are based on tradition model-based framework with multiple steps, showing poor efficiency and restricted performance. Thus, we propose an interferometric spectrum reconstruction method based on degradation synthesis and deep learning.Firstly, based on imaging mechanism, we proposed an mathematical model of interferometric imaging to analyse the degradation components as noises and trends during imaging. The model consists of three stages, namely instrument degradation, sensing degradation, and signal-independent degradation process. Then, we designed calibration-based method to estimate parameters in the model, of which the results are used for synthesizing realistic dataset for learning-based algorithms.In addition, we proposed a dual-stage interferogram spectrum reconstruction framework, which supports pre-training and integration of denoising DNNs. Experiments exhibits the reliability of our degradation model and synthesized data, and the effectiveness of the proposed reconstruction method.
文摘This article presents hourly load forecasting by using an incremental learning model called Online Sequential Extreme Learning Machine(OSELM),which can learn and adapt automatically according to new arrival input.However,the use of OS-ELM requires a sufficient amount of initial training sample data,which makes OS-ELM inoperable if sufficiently accurate sample data cannot be obtained.To solve this problem,a synthesis of the initial training sample is proposed.The synthesis of the initial sample is achieved by taking the first data received at the start of working and adding random noises to that data to create new and sufficient samples.Then the synthesis samples are used to initial train the OS-ELM.This proposed method is compared with Fully Online Extreme Learning Machine(FOS-ELM),which is an incremental learning model that also does not require the initial training samples.Both the proposed method and FOS-ELM are used for hourly load forecasting from the Hourly Energy Consumption dataset.Experiments have shown that the proposed method with a wide range of noise levels,can forecast hourly load more accurately than the FOS-ELM.
基金This work was partly supported by the ETH Zürich Fund(OK),and by Huawei grants.
文摘While recent years have witnessed a dramatic upsurge of exploiting deep neural networks toward solving image denoising,existing methods mostly rely on simple noise assumptions,such as additive white Gaussian noise(AWGN),JPEG compression noise and camera sensor noise,and a general-purpose blind denoising method for real images remains unsolved.In this paper,we attempt to solve this problem from the perspective of network architecture design and training data synthesis.Specifically,for the network architecture design,we propose a swin-conv block to incorporate the local modeling ability of residual convolutional layer and non-local modeling ability of swin transformer block,and then plug it as the main building block into the widely-used image-to-image translation UNet architecture.For the training data synthesis,we design a practical noise degradation model which takes into consideration different kinds of noise(including Gaussian,Poisson,speckle,JPEG compression,and processed camera sensor noises)and resizing,and also involves a random shuffle strategy and a double degradation strategy.Extensive experiments on AGWN removal and real image denoising demonstrate that the new network architecture design achieves state-of-the-art performance and the new degradation model can help to significantly improve the practicability.We believe our work can provide useful insights into current denoising research.The source code is available at https://github.com/cszn/SCUNet.
文摘Purpose:We aimed to perform a systematic review and meta-analysis of the effects of training to muscle failure or non-failure on muscular strength and hypertrophy.Methods:Meta-analyses of effect sizes(ESs)explored the effects of training to failure vs.non-failure on strength and hypertrophy.Subgroup meta-analyses explored potential moderating effects of variables such as training status(trained vs.untrained),training volume(volume equated vs.volume non-equated),body region(upper vs.lower),exercise selection(multi-vs.single-joint exercises(only for strength)),and study design(independent vs.dependent groups).Results:Fifteen studies were included in the review.All studies included young adults as participants.Meta-analysis indicated no significant difference between the training conditions for muscular strength(ES=-0.09,95%confidence interval(95%CI):-0.22 to 0.05)and for hypertrophy(ES=0.22,95%CI:-0.11 to 0.55).Subgroup analyses that stratified the studies according to body region,exercise selection,or study design showed no significant differences between training conditions.In studies that did not equate training volume between the groups,the analysis showed significant favoring of non-failure training on strength gains(ES=-0.32,95%CI:-0.57 to-0.07).In the subgroup analysis for resistance-trained individuals,the analysis showed a significant effect of training to failure for muscle hypertrophy(ES=0.15,95%CI:0.03-0.26).Conclusion:Training to muscle failure does not seem to be required for gains in strength and muscle size.However,training in this manner does not seem to have detrimental effects on these adaptations,either.More studies should be conducted among older adults and highly trained individuals to improve the generalizability of these findings.
文摘Purpose:This review aimed to synthesize previous findings on the test-retest reliability of the 30-15 Intermittent Fitness Test(IFT).Methods:The literature searches were performed in 8 databases.Studies that examined the test-retest reliability of the 30-15 IFT and presented the intraclass correlation coefficient(ICC) and/or the coefficient of variation(CV) for maximal velocity and/or peak heart rate were included.The consensus-based standards for the selection of health measurement instruments(COSMIN) checklist was used for the assessment of the methodological quality of the included studies.Results:Seven studies,with a total of 10 study groups,explored reliability of maximal velocity assessed by the 30-15 IFT.ICCs ranged from0.80 to 0.99,where 70% of ICCs were≥0.90.CVs for maximal velocity ranged from 1.5% to 6.0%.Six studies,with a total of 7 study groups,explored reliability of peak heart rate as assessed by the 30-15 IFT.ICCs ranged from 0.90 to 0.97(i.e.,all ICCs were≥0.90).CVs ranged from 0.6% to 4.8%.All included studies were of excellent methodological quality.Conclusion:From the results of this systematic review,it can be concluded that the 30-15 IFT has excellent test-retest reliability for both maximal velocity and peak heart rate.The test may,therefore,be used as a reliable measure of fitness in research and sports practice.
基金was supported by the funds of Bejing Advanced Innovation Center for Language Resources.(TYZ19005)Research Program of State Language Commission(ZDI135-105,YB135-89).
文摘Due to the lack of parallel data in current grammatical error correction(GEC)task,models based on sequence to sequence framework cannot be adequately trained to obtain higher performance.We propose two data synthesis methods which can control the error rate and the ratio of error types on synthetic data.The first approach is to corrupt each word in the monolingual corpus with a fixed probability,including replacement,insertion and deletion.Another approach is to train error generation models and further filtering the decoding results of the models.The experiments on different synthetic data show that the error rate is 40%and that the ratio of error types is the same can improve the model performance better.Finally,we synthesize about 100 million data and achieve comparable performance as the state of the art,which uses twice as much data as we use.
文摘The standard approach to tackling computer vision problems is to train deep convolutional neural network(CNN)models using large-scale image datasets that are representative of the target task.However,in many scenarios,it is often challenging to obtain sufficient image data for the target task.Data augmentation is a way to mitigate this challenge.A common practice is to explicitly transform existing images in desired ways to create the required volume and variability of training data necessary to achieve good generalization performance.In situations where data for the target domain are not accessible,a viable workaround is to synthesize training data from scratch,i.e.,synthetic data augmentation.This paper presents an extensive review of synthetic data augmentation techniques.It covers data synthesis approaches based on realistic 3D graphics modelling,neural style transfer(NST),differential neural rendering,and generative modelling using generative adversarial networks(GANs)and variational autoencoders(VAEs).For each of these classes of methods,we focus on the important data generation and augmentation techniques,general scope of application and specific use-cases,as well as existing limitations and possible workarounds.Additionally,we provide a summary of common synthetic datasets for training computer vision models,highlighting the main features,application domains and supported tasks.Finally,we discuss the effectiveness of synthetic data augmentation methods.Since this is the first paper to explore synthetic data augmentation methods in great detail,we are hoping to equip readers with the necessary background information and in-depth knowledge of existing methods and their attendant issues.
基金This work was supported by the 863 National High Technology Project and the National Natural Science Foundation of China (No. 60275014).
文摘Prosodic control is an important part of speech synthesis system. Prosodic parameters choice right or wrong influences the quality of synthetic speech directly. At present, text to speech system has less effective describe to reflect data relationships in the corpus. A new research approach - data mining technology to discover those relationships by association rules modeling is presented. And a new algorithm for generating association rules of prosodic parameters including pitch parameters and duration parameters from corpus is developed. The output rules improve the correctness of syllable choice in text to speech system.
文摘Classification of sheep behaviour from a sequence of tri-axial accelerometer data has the potential to enhance sheep management.Sheep behaviour is inherently imbalanced(e.g.,more ruminating than walking)resulting in underperforming classification for the minority activities which hold importance.Existing works have not addressed class imbalance and use traditional machine learning techniques,e.g.,Random Forest(RF).We investigated Deep Learning(DL)models,namely,Long Short Term Memory(LSTM)and Bidirectional LSTM(BLSTM),appropriate for sequential data,from imbalanced data.Two data sets were collected in normal grazing conditions using jaw-mounted and earmounted sensors.Novel to this study,alongside typical single classes,e.g.,walking,depending on the behaviours,data samples were labelled with compound classes,e.g.,walking_-grazing.The number of steps a sheep performed in the observed 10 s time window was also recorded and incorporated in the models.We designed several multi-class classification studies with imbalance being addressed using synthetic data.DL models achieved superior performance to traditional ML models,especially with augmented data(e.g.,4-Class+Steps:LSTM 88.0%,RF 82.5%).DL methods showed superior generalisability on unseen sheep(i.e.,F1-score:BLSTM 0.84,LSTM 0.83,RF 0.65).LSTM,BLSTM and RF achieved sub-millisecond average inference time,making them suitable for real-time applications.The results demonstrate the effectiveness of DL models for sheep behaviour classification in grazing conditions.The results also demonstrate the DL techniques can generalise across different sheep.The study presents a strong foundation of the development of such models for real-time animal monitoring.
文摘Area and test time are two major overheads encountered duringdata path high level synthesis for BIST. This paper presents an approach to behavioral synthesis for loop-based BIST. By taking into account the requirements of theBIST scheme during behavioral synthesis processes, an area optimal BIST solutioncan be obtained. This approach is based on the use of test resources reusabilitythat results in a fewer number of registers being modified to be test registers. Thisis achieved by incorporating self-testability constraints during register assignmentoperations. Experimental results on benchmarks are presented to demonstrate theeffectiveness of the approach.
