The demand for adopting neural networks in resource-constrained embedded devices is continuously increasing.Quantization is one of the most promising solutions to reduce computational cost and memory storage on embedd...The demand for adopting neural networks in resource-constrained embedded devices is continuously increasing.Quantization is one of the most promising solutions to reduce computational cost and memory storage on embedded devices.In order to reduce the complexity and overhead of deploying neural networks on Integeronly hardware,most current quantization methods use a symmetric quantization mapping strategy to quantize a floating-point neural network into an integer network.However,although symmetric quantization has the advantage of easier implementation,it is sub-optimal for cases where the range could be skewed and not symmetric.This often comes at the cost of lower accuracy.This paper proposed an activation redistribution-based hybrid asymmetric quantizationmethod for neural networks.The proposedmethod takes data distribution into consideration and can resolve the contradiction between the quantization accuracy and the ease of implementation,balance the trade-off between clipping range and quantization resolution,and thus improve the accuracy of the quantized neural network.The experimental results indicate that the accuracy of the proposed method is 2.02%and 5.52%higher than the traditional symmetric quantization method for classification and detection tasks,respectively.The proposed method paves the way for computationally intensive neural network models to be deployed on devices with limited computing resources.Codes will be available on https://github.com/ycjcy/Hybrid-Asymmetric-Quantization.展开更多
The quantization algorithm compresses the original network by reducing the numerical bit width of the model,which improves the computation speed. Because different layers have different redundancy and sensitivity to d...The quantization algorithm compresses the original network by reducing the numerical bit width of the model,which improves the computation speed. Because different layers have different redundancy and sensitivity to databit width. Reducing the data bit width will result in a loss of accuracy. Therefore, it is difficult to determinethe optimal bit width for different parts of the network with guaranteed accuracy. Mixed precision quantizationcan effectively reduce the amount of computation while keeping the model accuracy basically unchanged. In thispaper, a hardware-aware mixed precision quantization strategy optimal assignment algorithm adapted to low bitwidth is proposed, and reinforcement learning is used to automatically predict the mixed precision that meets theconstraints of hardware resources. In the state-space design, the standard deviation of weights is used to measurethe distribution difference of data, the execution speed feedback of simulated neural network accelerator inferenceis used as the environment to limit the action space of the agent, and the accuracy of the quantization model afterretraining is used as the reward function to guide the agent to carry out deep reinforcement learning training. Theexperimental results show that the proposed method obtains a suitable model layer-by-layer quantization strategyunder the condition that the computational resources are satisfied, and themodel accuracy is effectively improved.The proposed method has strong intelligence and certain universality and has strong application potential in thefield of mixed precision quantization and embedded neural network model deployment.展开更多
[Objectives]This study was conducted to obtain the best solution for sterilizing tall fescue(Festuca arundinacea)and ryegrass(Lolium perenne)seeds by heating in a water bath for a short time.[Methods]The tall fescue a...[Objectives]This study was conducted to obtain the best solution for sterilizing tall fescue(Festuca arundinacea)and ryegrass(Lolium perenne)seeds by heating in a water bath for a short time.[Methods]The tall fescue and ryegrass seeds infected with endophytic fungi were sterilized by heating in a water bath to compare seed germination and initial growth of turfgrass seedlings under different treatment time.[Results]Sterilization in a 60℃water bath for 20 and 30 min both inhibited the germination of tall fescue and ryegrass seeds,while the 20 min treatment did not significantly affect the lengths of stems,leaves and roots of seedlings.The 20 min water bath sterilization treatment had no effects on the plant heights and biomass of the two turfgrass seedlings and the tiller number of ryegrass,but sterilization for 30 min significantly reduced the aboveground and total biomass of seedlings and the tiller number of ryegrass.[Conclusions]Sterilization in a water bath at 60℃ for 20 min achieved rapid sterilization in a short time,without significantly negatively affecting the growth of seedlings.展开更多
Persistent asymptomatic(PA)SARS-CoV-2 infections have been identified.The immune responses in these patients are unclear,and the development of effective treatments for these patients is needed.Here,we report a cohort...Persistent asymptomatic(PA)SARS-CoV-2 infections have been identified.The immune responses in these patients are unclear,and the development of effective treatments for these patients is needed.Here,we report a cohort of 23 PA cases carrying viral RNA for up to 191 days.PA cases displayed low levels of inflammatory and interferon response,weak antibody response,diminished circulating follicular helper T cells(cTfh),and inadequate specific CD4+and CD8+T-cell responses during infection,which is distinct from symptomatic infections and resembling impaired immune activation.Administration of a single dose of Ad5-nCoV vaccine to 10 of these PA cases elicited rapid and robust antibody responses as well as coordinated B-cell and cTfh responses,resulting in successful viral clearance.Vaccine-induced antibodies were able to neutralize various variants of concern and persisted for over 6 months,indicating long-term protection.Therefore,our study provides an insight into the immune status of PA infections and highlights vaccination as a potential treatment for prolonged SARS-CoV-2 infections.展开更多
基金The Qian Xuesen Youth Innovation Foundation from China Aerospace Science and Technology Corporation(Grant Number 2022JY51).
文摘The demand for adopting neural networks in resource-constrained embedded devices is continuously increasing.Quantization is one of the most promising solutions to reduce computational cost and memory storage on embedded devices.In order to reduce the complexity and overhead of deploying neural networks on Integeronly hardware,most current quantization methods use a symmetric quantization mapping strategy to quantize a floating-point neural network into an integer network.However,although symmetric quantization has the advantage of easier implementation,it is sub-optimal for cases where the range could be skewed and not symmetric.This often comes at the cost of lower accuracy.This paper proposed an activation redistribution-based hybrid asymmetric quantizationmethod for neural networks.The proposedmethod takes data distribution into consideration and can resolve the contradiction between the quantization accuracy and the ease of implementation,balance the trade-off between clipping range and quantization resolution,and thus improve the accuracy of the quantized neural network.The experimental results indicate that the accuracy of the proposed method is 2.02%and 5.52%higher than the traditional symmetric quantization method for classification and detection tasks,respectively.The proposed method paves the way for computationally intensive neural network models to be deployed on devices with limited computing resources.Codes will be available on https://github.com/ycjcy/Hybrid-Asymmetric-Quantization.
文摘The quantization algorithm compresses the original network by reducing the numerical bit width of the model,which improves the computation speed. Because different layers have different redundancy and sensitivity to databit width. Reducing the data bit width will result in a loss of accuracy. Therefore, it is difficult to determinethe optimal bit width for different parts of the network with guaranteed accuracy. Mixed precision quantizationcan effectively reduce the amount of computation while keeping the model accuracy basically unchanged. In thispaper, a hardware-aware mixed precision quantization strategy optimal assignment algorithm adapted to low bitwidth is proposed, and reinforcement learning is used to automatically predict the mixed precision that meets theconstraints of hardware resources. In the state-space design, the standard deviation of weights is used to measurethe distribution difference of data, the execution speed feedback of simulated neural network accelerator inferenceis used as the environment to limit the action space of the agent, and the accuracy of the quantization model afterretraining is used as the reward function to guide the agent to carry out deep reinforcement learning training. Theexperimental results show that the proposed method obtains a suitable model layer-by-layer quantization strategyunder the condition that the computational resources are satisfied, and themodel accuracy is effectively improved.The proposed method has strong intelligence and certain universality and has strong application potential in thefield of mixed precision quantization and embedded neural network model deployment.
基金the National Natural Science Foundation of China(31870484).
文摘[Objectives]This study was conducted to obtain the best solution for sterilizing tall fescue(Festuca arundinacea)and ryegrass(Lolium perenne)seeds by heating in a water bath for a short time.[Methods]The tall fescue and ryegrass seeds infected with endophytic fungi were sterilized by heating in a water bath to compare seed germination and initial growth of turfgrass seedlings under different treatment time.[Results]Sterilization in a 60℃water bath for 20 and 30 min both inhibited the germination of tall fescue and ryegrass seeds,while the 20 min treatment did not significantly affect the lengths of stems,leaves and roots of seedlings.The 20 min water bath sterilization treatment had no effects on the plant heights and biomass of the two turfgrass seedlings and the tiller number of ryegrass,but sterilization for 30 min significantly reduced the aboveground and total biomass of seedlings and the tiller number of ryegrass.[Conclusions]Sterilization in a water bath at 60℃ for 20 min achieved rapid sterilization in a short time,without significantly negatively affecting the growth of seedlings.
基金the National Key R&D Program of China(2021YFC2302004 and 2021YFC2301102)National Natural Science Foundation of China(82172240,82101836,81901623 and 82201933)+6 种基金China National Postdoctoral Program for Innovative Talents(BX2021087),China Postdoctoral Science Foundation(2021M690787 and 2022M710892)Guangdong Basic and Applied Basic Research Foundation(2021B1515130005 and 2022B1515020059)Guangdong Medical Technology Research Foundation(B2022233)Guangzhou Science and Technology Planning Program Municipal School/Institute-Joint Funded Program(202102010143)PhD Startup Foundation from Guangzhou Women and Children’s Medical Center(2018-2020)State Key Laboratory of Respiratory Diseases Funded Program(SKLRD-Z-202330)Zhong Nanshan Medical Foundation of Guangdong Province(ZNSA-202001,ZNSXS-2020012,ZNSA-2020013 and ZNSXS-20220015).
文摘Persistent asymptomatic(PA)SARS-CoV-2 infections have been identified.The immune responses in these patients are unclear,and the development of effective treatments for these patients is needed.Here,we report a cohort of 23 PA cases carrying viral RNA for up to 191 days.PA cases displayed low levels of inflammatory and interferon response,weak antibody response,diminished circulating follicular helper T cells(cTfh),and inadequate specific CD4+and CD8+T-cell responses during infection,which is distinct from symptomatic infections and resembling impaired immune activation.Administration of a single dose of Ad5-nCoV vaccine to 10 of these PA cases elicited rapid and robust antibody responses as well as coordinated B-cell and cTfh responses,resulting in successful viral clearance.Vaccine-induced antibodies were able to neutralize various variants of concern and persisted for over 6 months,indicating long-term protection.Therefore,our study provides an insight into the immune status of PA infections and highlights vaccination as a potential treatment for prolonged SARS-CoV-2 infections.
