[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-125623-en":3,"doc-seo-125623-105":30,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":4,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":27,"seo_description":14,"update_tm":28,"read_time":29},125623,549758146520,"Patrick","https://ap-avatar.wpscdn.com/avatar/80002397d8c0411e94?_k=1775819394049821470",8,"Research & Report","FEW-SHOT IMAGE CLASSIFICATION BASED ON GRADUAL MACHINE LEARNING - Technical Report","Few-shot image classification focuses on assigning accurate labels to unlabeled images using only a small number of labeled samples. Deep learning methods often rely on complex backbones, yet remain difficult when knowledge from training classes must transfer to new ones. This report presents a gradual machine learning approach under a non-i.i.d paradigm that progressively labels target images by increasing hardness via iterative factor inference in a factor graph. Unary factors use class center distance in an embedding space, and binary factors use k-nearest neighborhood. Experiments on benchmark datasets show 1–5% accuracy improvements over state of the art, with performance that keeps improving as the query set grows while conventional deep models plateau or degrade.","arXiv :2307 . 15524v1 [ cs .CV] 28 Jul 2023  \nFEW-SHOT IMAGE CLASSIFICATION BASED ON GRADUAL  \nMACHINE LEARNING  \nTECHNICAL REPORT  \nNa Chen1 , Xianming Kuang2 , Feiyu Liu1 , Kehao Wang2 , Qun Chen1,2 ∗  \n1 School of software,Northwestern Polytechnical University,Xi’an, China, 710072  \n2 School of Computer Science,Northwestern Polytechnical University,Xi’an, China, 710072  \nJuly 31, 2023  \nABSTRACT  \nFew-shot image classification aims to accurately classify unlabeled images using only a few labeled samples. The state-of-the-art solutions are built by deep learning, which focuses on designing increasingly complex deep backbones. Unfortunately, the task remains very challenging due to the difficulty of transferring the knowledge learned in training classes to new ones. In this paper, we propose a novel approach based on the non-i.i.d paradigm of gradual machine learning (GML) . It begins with only a few labeled observations, and then gradually labels target images in the increasing order of hardness by iterative factor inference in a factor graph. Specifically, our proposed solution extracts indicative feature representations by deep backbones, and then constructs both unaryand binary factors based on the extracted features to facilitate gradual learning. The unary factors are constructed based on class center distance in an embedding space, while the binary factors are constructed based on k-nearest neighborhood. We have empirically validated the performance of the proposed approach on benchmark datasets by a comparative study. Our extensive experiments demonstrate that the proposed approach can improve the SOTA performance by 1-5% in terms of accuracy. More notably, it is more robust than the existing deep models in that its performance can consistently improve as the size of query set increases while the performance of deep models remains essentially flat or even becomes worse.  \nKeywords Few-shot Image Classification · gradual machine learning · Factor Graph  \n1 Introduction  \nExtensively studied in the literature, image classification can be accurately performed by Deep Neural Network (DNN) models provided that there are sufficient labeled training data [1, 2] . Unfortunately, in many application scenarios (e.g., medical image analysis [3] and autonomous driving [4]), large amounts of labeled images may not be readily available because data acquisition and annotation needs to involve intensive manual effort. Under these circumstances, DNN models can easily overfit and fail to achieve satisfactory performance. To address this limitation, few-shot image classification has been proposed to classify unseen classes with only a few labeled samples [5, 6] .  \nThe existing approaches for few-shot image classification can be broadly categorized into two groups: inductive fewshot learning and transductive few-shot learning. Inductive learning typically trains a generic model based on the labeled samples in training classes, and then directly uses the learnt model to classify each unlabeled sample in test classes independently from each other [7, 8, 9, 10, 11, 12] . In contrast, supposing that it has the access to both labeled and unlabeled samples in test classes, transductive learning performs class label inference jointly for all the unlabeled samples.  \n∗[Joint corresponding author:chenbenben@nwpu.edu.cn](Joint corresponding author:chenbenben@nwpu.edu.cn)  \nEasy Instance Labeling  \nFeature Extraction &  \nInfluence Modeling  \n!!  \n\"!  \nfeature value  \n1  \n􀀢 !  \n􀀢 !  \n􀀢\"  \n􀀢& 􀀢\\#  \n􀀢\\# 􀀢$ 􀀢%  \n􀀢 !  \n􀀢\"  \n􀀢\\#'!  \n􀀢\\#'*  \n􀀢$(!􀀢$  \n􀀢\\#  \n􀀢%  \n􀀢)  \n!!  \n\"!  \nfeature value  \n1  \nFactor Graph  \nGradual Inference 􀀢 ! 􀀢\" 􀀢\\# 􀀢$ 􀀢% 􀀢\\#  \nlabeled images  unlabeled images  \n unary features(􀀣􀀣􀀤)  \n binary features(􀀥􀀦􀀦)  \nFigure 1: The framework of GML for few-shot image classification: there are two types of factors in the constructed factor graph, which correspond to the unary class centroid distance (CCD) and binary k-nearest neighborhood (KN","cbCaigPC5LWd5rO2","https://ap.wps.com/l/cbCaigPC5LWd5rO2","pdf",3328754,1,17,"English","en",105,"# Introduction\n## Problem background and motivation\n## Inductive vs transductive few-shot learning\n## Proposed non-i.i.d gradual machine learning approach\n## Factor graph framework and feature modeling","[{\"question\":\"What is the main goal of few-shot image classification in this work?\",\"answer\":\"To classify unlabeled images accurately using only a few labeled samples, especially when transferring knowledge to new classes with scarce annotations.\"},{\"question\":\"How does the proposed method perform gradual learning?\",\"answer\":\"It gradually labels query images in increasing order of hardness using iterative factor inference in a factor graph, starting from the few labeled support samples.\"},{\"question\":\"What do the unary and binary factors represent in the factor graph?\",\"answer\":\"Unary factors are built from class center distance in an embedding space, while binary factors are built from k-nearest neighborhood relationships.\"}]","FEW-SHOT IMAGE CLASSIFICATION BASED ON GRADUAL MACHINE LEARNING - Technical Report | PDF",1785900275,43,{"code":4,"msg":31,"data":32},"ok",{"site_id":24,"language":23,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"few-shot-image-classification-based-on-gradual-machine-learning-technical-report","",{"@graph":36,"@context":85},[37,54,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,48,51],{"item":41,"name":42,"@type":43,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":47},"https://docshare.wps.com/document/","Document",2,{"item":49,"name":12,"@type":43,"position":50},"https://docshare.wps.com/document/research-report/",3,{"item":52,"name":13,"@type":43,"position":53},"https://docshare.wps.com/document/few-shot-image-classification-based-on-gradual-machine-learning-technical-report/125623/",4,{"url":52,"name":13,"@type":55,"author":56,"headline":13,"publisher":58,"fileFormat":61,"inLanguage":23,"description":14,"dateModified":62,"datePublished":62,"encodingFormat":61,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":57},"Person",{"url":41,"name":59,"@type":60},"DocShare","Organization","application/pdf","2026-08-05",true,{"@type":65,"interactionType":66,"userInteractionCount":4},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What is the main goal of few-shot image classification in this work?","Question",{"text":75,"@type":76},"To classify unlabeled images accurately using only a few labeled samples, especially when transferring knowledge to new classes with scarce annotations.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does the proposed method perform gradual learning?",{"text":80,"@type":76},"It gradually labels query images in increasing order of hardness using iterative factor inference in a factor graph, starting from the few labeled support samples.",{"name":82,"@type":73,"acceptedAnswer":83},"What do the unary and binary factors represent in the factor graph?",{"text":84,"@type":76},"Unary factors are built from class center distance in an embedding space, while binary factors are built from k-nearest neighborhood relationships.","https://schema.org",{"og:url":52,"og:type":87,"og:title":13,"og:site_name":59,"og:description":14},"article",{"robots":89,"canonical":52},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":47,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":53,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]