The majority of existing graph-network-based few-shot models focus on a node-similarity update mode.The lack of adequate information intensies the risk of overtraining.In this paper,we propose a novel Multihead Attent...The majority of existing graph-network-based few-shot models focus on a node-similarity update mode.The lack of adequate information intensies the risk of overtraining.In this paper,we propose a novel Multihead Attention Graph Network to excavate discriminative relation and fulll effective information propagation.For edge update,the node-level attention is used to evaluate the similarities between the two nodes and the distributionlevel attention extracts more in-deep global relation.The cooperation between those two parts provides a discriminative and comprehensive expression for edge feature.For node update,we embrace the label-level attention to soften the noise of irrelevant nodes and optimize the update direction.Our proposed model is veried through extensive experiments on two few-shot benchmark MiniImageNet and CIFAR-FS dataset.The results suggest that our method has a strong capability of noise immunity and quick convergence.The classication accuracy outperforms most state-of-the-art approaches.展开更多
Fine-tuning is a popular approach to solve the few-shot object detection problem.In this paper,we attempt to introduce a new perspective on it.We formulate the few-shot novel tasks as a type of distribution shifted fr...Fine-tuning is a popular approach to solve the few-shot object detection problem.In this paper,we attempt to introduce a new perspective on it.We formulate the few-shot novel tasks as a type of distribution shifted from its ground-truth distribution.We introduce the concept of imaginary placeholder masks to show that this distribution shift is essentially a composite of in-distribution(ID)and out-of-distribution(OOD)shifts.Our empirical investigation results show that it is significant to balance the trade-off between adapting to the available few-shot distribution and keeping the distribution-shift robustness of the pre-trained model.We explore improvements in the few-shot finetuning transfer in the few-shot object detection(FSOD)settings from three aspects.First,we explore the LinearProbe-Finetuning(LP-FT)technique to balance this trade-off to mitigate the feature distortion problem.Second,we explore the effectiveness of utilizing the protection freezing strategy for querybased object detectors to keep their OOD robustness.Third,we try to utilize ensembling methods to circumvent the feature distortion.All these techniques are integrated into a whole method called BIOT(Balanced ID-OOD Transfer).Evaluation results show that our method is simple yet effective and general to tap the FSOD potential of query-based object detectors.It outperforms the current SOTA method in many FSOD settings and has a promising scaling capability.展开更多
Assessing wear is an indispensable task across almost all engineering disciplines,and automated wear assessment would be highly desirable.To determine the occurrence of wear,machine learning strategies have already be...Assessing wear is an indispensable task across almost all engineering disciplines,and automated wear assessment would be highly desirable.To determine the occurrence of wear,machine learning strategies have already been successfully applied.However,classifying different types of wear remains challenging.Additionally,data scarcity is a major bottle neck that limits the applicability of machine learning models in certain areas such as biomedical engineering.Here,we present a method to accurately classify surface topographies representing the three most common types of mechanically induced wear:abrasive,erosive,and adhesive wear.First,a random forest(RF)classifier is trained on a list of parameters determined from 3-dimensional(3D)surface scans.Then,this method is adapted to a small dataset obtained from damaged cartilage tissue by using knowledge transfer principles.In detail,two random forest models are trained separately:a base model on a large training dataset obtained on synthetic samples,and a complementary model on the scarce cartilage data.After the separate training phases,the decision trees of both models are combined for inference on the scarce cartilage data.This model architecture provides a highly adaptable framework for assessing wear on biological samples and requires only a handful of training data.A similar approach might also be useful in many other areas of materials science where training data are difficult to obtain.展开更多
基金supported in part by the Natural Science Foundation of China under Grant 61972169 and U1536203in part by the National key research and developm program of China(2016QY01W0200)in part by the Major Scientic and Technological Project of Hubei Province(2018AAA068 and 2019AAA051).
文摘The majority of existing graph-network-based few-shot models focus on a node-similarity update mode.The lack of adequate information intensies the risk of overtraining.In this paper,we propose a novel Multihead Attention Graph Network to excavate discriminative relation and fulll effective information propagation.For edge update,the node-level attention is used to evaluate the similarities between the two nodes and the distributionlevel attention extracts more in-deep global relation.The cooperation between those two parts provides a discriminative and comprehensive expression for edge feature.For node update,we embrace the label-level attention to soften the noise of irrelevant nodes and optimize the update direction.Our proposed model is veried through extensive experiments on two few-shot benchmark MiniImageNet and CIFAR-FS dataset.The results suggest that our method has a strong capability of noise immunity and quick convergence.The classication accuracy outperforms most state-of-the-art approaches.
文摘Fine-tuning is a popular approach to solve the few-shot object detection problem.In this paper,we attempt to introduce a new perspective on it.We formulate the few-shot novel tasks as a type of distribution shifted from its ground-truth distribution.We introduce the concept of imaginary placeholder masks to show that this distribution shift is essentially a composite of in-distribution(ID)and out-of-distribution(OOD)shifts.Our empirical investigation results show that it is significant to balance the trade-off between adapting to the available few-shot distribution and keeping the distribution-shift robustness of the pre-trained model.We explore improvements in the few-shot finetuning transfer in the few-shot object detection(FSOD)settings from three aspects.First,we explore the LinearProbe-Finetuning(LP-FT)technique to balance this trade-off to mitigate the feature distortion problem.Second,we explore the effectiveness of utilizing the protection freezing strategy for querybased object detectors to keep their OOD robustness.Third,we try to utilize ensembling methods to circumvent the feature distortion.All these techniques are integrated into a whole method called BIOT(Balanced ID-OOD Transfer).Evaluation results show that our method is simple yet effective and general to tap the FSOD potential of query-based object detectors.It outperforms the current SOTA method in many FSOD settings and has a promising scaling capability.
文摘Assessing wear is an indispensable task across almost all engineering disciplines,and automated wear assessment would be highly desirable.To determine the occurrence of wear,machine learning strategies have already been successfully applied.However,classifying different types of wear remains challenging.Additionally,data scarcity is a major bottle neck that limits the applicability of machine learning models in certain areas such as biomedical engineering.Here,we present a method to accurately classify surface topographies representing the three most common types of mechanically induced wear:abrasive,erosive,and adhesive wear.First,a random forest(RF)classifier is trained on a list of parameters determined from 3-dimensional(3D)surface scans.Then,this method is adapted to a small dataset obtained from damaged cartilage tissue by using knowledge transfer principles.In detail,two random forest models are trained separately:a base model on a large training dataset obtained on synthetic samples,and a complementary model on the scarce cartilage data.After the separate training phases,the decision trees of both models are combined for inference on the scarce cartilage data.This model architecture provides a highly adaptable framework for assessing wear on biological samples and requires only a handful of training data.A similar approach might also be useful in many other areas of materials science where training data are difficult to obtain.
