Knowledge distillation(KD) enhances student network generalization by transferring dark knowledge from a complex teacher network. To optimize computational expenditure and memory utilization, self-knowledge distillati...Knowledge distillation(KD) enhances student network generalization by transferring dark knowledge from a complex teacher network. To optimize computational expenditure and memory utilization, self-knowledge distillation(SKD) extracts dark knowledge from the model itself rather than an external teacher network. However, previous SKD methods performed distillation indiscriminately on full datasets, overlooking the analysis of representative samples. In this work, we present a novel two-stage approach to providing targeted knowledge on specific samples, named two-stage approach self-knowledge distillation(TOAST). We first soften the hard targets using class medoids generated based on logit vectors per class. Then, we iteratively distill the under-trained data with past predictions of half the batch size. The two-stage knowledge is linearly combined, efficiently enhancing model performance. Extensive experiments conducted on five backbone architectures show our method is model-agnostic and achieves the best generalization performance.Besides, TOAST is strongly compatible with existing augmentation-based regularization methods. Our method also obtains a speedup of up to 2.95x compared with a recent state-of-the-art method.展开更多
Fruits,as the reproductive organs of many higher plants,are an integral part of a balanced diet,providing rich nutrients and bioactive substances for our health.Over the years,the research on fruit development,quality...Fruits,as the reproductive organs of many higher plants,are an integral part of a balanced diet,providing rich nutrients and bioactive substances for our health.Over the years,the research on fruit development,quality formation and stress response has deepened,which sheds light on the regulatory mechanism and application of fruit quality improvement.Among the many regulatory factors,microRNAs(miRNAs)are a class of short non-coding RNAs,ranging from 20 to 24-nt,which post-transcriptionally regulate target gene expression.miRNAs and their functions have been extensively examined in plants especially in model species,and they play critical roles in the regulation of diverse biological processes in response to endogenous developmental signals and external environmental cues,respectively.In fruit crops,the function of miRNAs and their regulation have also been under intensive study.In particular,the novel roles of miRNAs that have not been revealed in the model annual species have been unraveling,which reflect the genetic,physiological,and developmental complexity of gene regulation in fruit crops.Here we review the current research progress achieved,specifically in fruit crops,with a focus on the function of miRNAs in the control of fruit development and quality as well as response to various stresses.The future prospects of miRNAs for quality-targeted fruit breeding are also discussed.展开更多
Designing and preparing high-performance lignin-based dispersants are crucial steps in realizing the value-added utilization of lignin on an industrial scale.Such process depends heavily on an understanding of the dis...Designing and preparing high-performance lignin-based dispersants are crucial steps in realizing the value-added utilization of lignin on an industrial scale.Such process depends heavily on an understanding of the dispersion mechanism of lignin-based dispersants.Here,atomic force microscopy(AFM)is employed to quantitatively investigate the dispersion mechanism of a lignosulfonate/silica(LS/SiO_(2))system under different pH conditions.The results show that the repulsive force between SiO_(2)particles in LS solution is stronger than it is in water,resulting in better dispersion stability.The Derjaguin–Landau–Verwey–Overbeek(DLVO)formula as well as the DLVO formula combined with steric repulsion is utilized for the fitting of the AFM force/distance(F/D)curves between the SiO_(2)probe and substrate in water and in LS solution.Based on these fitting results,electrostatic and steric repulsive forces are respectively calculated,yielding further evidence that LS provides strong steric repulsion between SiO_(2)particles.Further studies indicate that the adsorbance of LS on SiO_(2)(Q),the normalized interaction constant(A),and the characteristic length(L)are the three critical factors affecting steric repulsion in the LS/SiO_(2)system.Based on the above conclusions,a novel quaternized grafted-sulfonation lignin(QAGSL)dispersant is designed and prepared.The QAGSL dispersant exhibits good dispersing performance for SiO_(2)and real cement particles.This work provides a fundamental and quantitative understanding of the dispersion mechanism in the LS/inorganic particle system and provides important guidance for the development of high-performance lignin-based dispersants.展开更多
The bandwidth of internet connections is still a bottleneck when transmitting large amounts of images,making the image quality assessment essential.Neurophysiological assessment of image quality has highlight advantag...The bandwidth of internet connections is still a bottleneck when transmitting large amounts of images,making the image quality assessment essential.Neurophysiological assessment of image quality has highlight advantages for it does not interfere with natural viewing behavior.However,in JPEG compression,the previous study is hard to tell the difference between the electroencephalogram(EEG)evoked by different quality images.In this paper,we propose an EEG analysis approach based on algebraic topology analysis,and the result shows that the difference between Euler characteristics of EEG evoked by different distortion images is striking both in the alpha and beta band.Moreover,we further discuss the relationship between the images and the EEG signals,and the results implied that the algebraic topological properties of images are consistent with that of brain perception,which is possible to give birth to braininspired image compression based on algebraic topological features.In general,an algebraic topologybased approach was proposed in this paper to analyze the perceptual characteristics of image quality,which will be beneficial to provide a reliable score for data compression in the network and improve the network transmission capacity.展开更多
To improve the performance of real-time recombinase polymerase amplification(RPA),a microfluidic system with active mixing is developed to optimize the reaction dynamics.Instead of adopting a single typical reaction c...To improve the performance of real-time recombinase polymerase amplification(RPA),a microfluidic system with active mixing is developed to optimize the reaction dynamics.Instead of adopting a single typical reaction chamber,a specific reactor including a relatively large chamber in center with two adjacent zig-zag channels at two sides is integrated into the microfluidic chip.Active mixing is achieved by driving the viscous reagent between the chamber and the channel back and forth periodically with an outside compact peristaltic pump.To avoid reagent evapora-tion,one end of the reactor is sealed with paraffin oil.A hand-held companion device is developed to facilitate real-time RPA amplification within 20 min.The whole area of the reactor is heated with a resistance heater to provide uniform reaction temperature.To achieve real-time monitoring,a compact fluorescence detection module is integrated into the hand-held device.A smartphone with custom application software is adopted to control the hand-held device and display the real-time fluorescence curves.The performances of two cases with and without active on-chip mixing are compared between each other by detecting African swine fever viruses.It has been demonstrated that,with active on-chip mixing,the amplification efficiency and detection sensitivity can be signifi-cantly improved.展开更多
ATP is the primary form of energy for plants,and a shortage of cellular ATP is generally acknowledged to pose a threat to plant growth and development,stress resistance,and crop quality.The overall metabolic processes...ATP is the primary form of energy for plants,and a shortage of cellular ATP is generally acknowledged to pose a threat to plant growth and development,stress resistance,and crop quality.The overall metabolic processes that contribute to the ATP pool,from production,dissipation,and transport to elimination,have been studied extensively.Considerable evidence has revealed that in addition to its role in energy supply,ATP also acts as a regulatory signaling molecule to activate global metabolic responses.Identification of the eATP receptor DORN1 contributed to a better understanding of how plants cope with disruption of ATP homeostasis and of the key points at which ATP signaling pathways intersect in cells or whole organisms.The functions of SnRK1a,the master regulator of the energy management network,in restoring the equi-librium of the ATP pool have been demonstrated,and the vast and complex metabolic network mediated by SnRK1a to adapt tofluctuating environments has been characterized.This paper reviews recent advances in understanding the regulatory control of the cellular ATP pool and discusses possible interactions among key regulators of ATP-pool homeostasis and crosstalk between iATP/eATP signaling pathways.Perception of ATP deficit and modulation of cellular ATP homeostasis mediated by SnRK1a in plants are discussed at the physiological and molecular levels.Finally,we suggest future research directions for modulation of plant cellular ATP homeostasis.展开更多
基金supported by the National Natural Science Foundation of China (62176061)。
文摘Knowledge distillation(KD) enhances student network generalization by transferring dark knowledge from a complex teacher network. To optimize computational expenditure and memory utilization, self-knowledge distillation(SKD) extracts dark knowledge from the model itself rather than an external teacher network. However, previous SKD methods performed distillation indiscriminately on full datasets, overlooking the analysis of representative samples. In this work, we present a novel two-stage approach to providing targeted knowledge on specific samples, named two-stage approach self-knowledge distillation(TOAST). We first soften the hard targets using class medoids generated based on logit vectors per class. Then, we iteratively distill the under-trained data with past predictions of half the batch size. The two-stage knowledge is linearly combined, efficiently enhancing model performance. Extensive experiments conducted on five backbone architectures show our method is model-agnostic and achieves the best generalization performance.Besides, TOAST is strongly compatible with existing augmentation-based regularization methods. Our method also obtains a speedup of up to 2.95x compared with a recent state-of-the-art method.
基金This work was supported by the National Natural Science Foundation of China(No.22273070 and No.92261101)the Innovation Capability Support Program of Shaanxi Province(2023-CX-TD-49)。
基金funded by the National Natural Science Foundation of China(31772371,32071810)Natural Science Foundation of Guangdong Province(2021A1515011258)。
文摘Fruits,as the reproductive organs of many higher plants,are an integral part of a balanced diet,providing rich nutrients and bioactive substances for our health.Over the years,the research on fruit development,quality formation and stress response has deepened,which sheds light on the regulatory mechanism and application of fruit quality improvement.Among the many regulatory factors,microRNAs(miRNAs)are a class of short non-coding RNAs,ranging from 20 to 24-nt,which post-transcriptionally regulate target gene expression.miRNAs and their functions have been extensively examined in plants especially in model species,and they play critical roles in the regulation of diverse biological processes in response to endogenous developmental signals and external environmental cues,respectively.In fruit crops,the function of miRNAs and their regulation have also been under intensive study.In particular,the novel roles of miRNAs that have not been revealed in the model annual species have been unraveling,which reflect the genetic,physiological,and developmental complexity of gene regulation in fruit crops.Here we review the current research progress achieved,specifically in fruit crops,with a focus on the function of miRNAs in the control of fruit development and quality as well as response to various stresses.The future prospects of miRNAs for quality-targeted fruit breeding are also discussed.
基金We gratefully acknowledge the financial support from the National Natural Science Foundation of China(21606089,21878113,and 21878114)the Guangdong Province Science and Technology Research Project of China(2017B090903003)the Guangzhou Science and Technology Research Project of China(201704030126 and 201806010139).
文摘Designing and preparing high-performance lignin-based dispersants are crucial steps in realizing the value-added utilization of lignin on an industrial scale.Such process depends heavily on an understanding of the dispersion mechanism of lignin-based dispersants.Here,atomic force microscopy(AFM)is employed to quantitatively investigate the dispersion mechanism of a lignosulfonate/silica(LS/SiO_(2))system under different pH conditions.The results show that the repulsive force between SiO_(2)particles in LS solution is stronger than it is in water,resulting in better dispersion stability.The Derjaguin–Landau–Verwey–Overbeek(DLVO)formula as well as the DLVO formula combined with steric repulsion is utilized for the fitting of the AFM force/distance(F/D)curves between the SiO_(2)probe and substrate in water and in LS solution.Based on these fitting results,electrostatic and steric repulsive forces are respectively calculated,yielding further evidence that LS provides strong steric repulsion between SiO_(2)particles.Further studies indicate that the adsorbance of LS on SiO_(2)(Q),the normalized interaction constant(A),and the characteristic length(L)are the three critical factors affecting steric repulsion in the LS/SiO_(2)system.Based on the above conclusions,a novel quaternized grafted-sulfonation lignin(QAGSL)dispersant is designed and prepared.The QAGSL dispersant exhibits good dispersing performance for SiO_(2)and real cement particles.This work provides a fundamental and quantitative understanding of the dispersion mechanism in the LS/inorganic particle system and provides important guidance for the development of high-performance lignin-based dispersants.
基金supported by the Key Research and Development Program of Zhejiang Province(Grant No.2019C03138 and No.2019C01002)。
文摘The bandwidth of internet connections is still a bottleneck when transmitting large amounts of images,making the image quality assessment essential.Neurophysiological assessment of image quality has highlight advantages for it does not interfere with natural viewing behavior.However,in JPEG compression,the previous study is hard to tell the difference between the electroencephalogram(EEG)evoked by different quality images.In this paper,we propose an EEG analysis approach based on algebraic topology analysis,and the result shows that the difference between Euler characteristics of EEG evoked by different distortion images is striking both in the alpha and beta band.Moreover,we further discuss the relationship between the images and the EEG signals,and the results implied that the algebraic topological properties of images are consistent with that of brain perception,which is possible to give birth to braininspired image compression based on algebraic topological features.In general,an algebraic topologybased approach was proposed in this paper to analyze the perceptual characteristics of image quality,which will be beneficial to provide a reliable score for data compression in the network and improve the network transmission capacity.
基金supported by the National Natural Science Foundation of China(Nos.81871505,61971026)the Fundamental Research Fund for the Central Universities(No.XK1802-4)+1 种基金the National Science and Technology Major Project(No.2018ZX10732101-001-009)the Research Fund to the Top Scientific and Technological Innovation Team from Beijing University of Chemical Technology(No.buctylkjcx06).
文摘To improve the performance of real-time recombinase polymerase amplification(RPA),a microfluidic system with active mixing is developed to optimize the reaction dynamics.Instead of adopting a single typical reaction chamber,a specific reactor including a relatively large chamber in center with two adjacent zig-zag channels at two sides is integrated into the microfluidic chip.Active mixing is achieved by driving the viscous reagent between the chamber and the channel back and forth periodically with an outside compact peristaltic pump.To avoid reagent evapora-tion,one end of the reactor is sealed with paraffin oil.A hand-held companion device is developed to facilitate real-time RPA amplification within 20 min.The whole area of the reactor is heated with a resistance heater to provide uniform reaction temperature.To achieve real-time monitoring,a compact fluorescence detection module is integrated into the hand-held device.A smartphone with custom application software is adopted to control the hand-held device and display the real-time fluorescence curves.The performances of two cases with and without active on-chip mixing are compared between each other by detecting African swine fever viruses.It has been demonstrated that,with active on-chip mixing,the amplification efficiency and detection sensitivity can be signifi-cantly improved.
基金supported by the National Natural Science Foundation of China (nos.32371926,32101567,and 32071810).
文摘ATP is the primary form of energy for plants,and a shortage of cellular ATP is generally acknowledged to pose a threat to plant growth and development,stress resistance,and crop quality.The overall metabolic processes that contribute to the ATP pool,from production,dissipation,and transport to elimination,have been studied extensively.Considerable evidence has revealed that in addition to its role in energy supply,ATP also acts as a regulatory signaling molecule to activate global metabolic responses.Identification of the eATP receptor DORN1 contributed to a better understanding of how plants cope with disruption of ATP homeostasis and of the key points at which ATP signaling pathways intersect in cells or whole organisms.The functions of SnRK1a,the master regulator of the energy management network,in restoring the equi-librium of the ATP pool have been demonstrated,and the vast and complex metabolic network mediated by SnRK1a to adapt tofluctuating environments has been characterized.This paper reviews recent advances in understanding the regulatory control of the cellular ATP pool and discusses possible interactions among key regulators of ATP-pool homeostasis and crosstalk between iATP/eATP signaling pathways.Perception of ATP deficit and modulation of cellular ATP homeostasis mediated by SnRK1a in plants are discussed at the physiological and molecular levels.Finally,we suggest future research directions for modulation of plant cellular ATP homeostasis.
