爆肝160+篇！近四年小樣本學(xué)習(xí)（FSL）頂會(huì)論文分享，含2023最新
學(xué)姐又爆肝了！速來(lái)白嫖?。ǖ沁€是想要個(gè)贊）
這次分享的是近四年（2020-2023）各大頂會(huì)中的小樣本學(xué)習(xí)（FSL）論文，有160+篇，涵蓋了FSL三大類(lèi)方法：數(shù)據(jù)、模型、算法，以及FSL的應(yīng)用、技術(shù)、理論等領(lǐng)域。
由于論文數(shù)量太多，學(xué)姐就不一一分析總結(jié)了，建議大家收藏下慢慢研讀。
全部160+篇論文原文及開(kāi)源代碼學(xué)姐已打包，需要的同學(xué)看這里??????
掃碼添加小享，回復(fù)“小樣本160”
免費(fèi)獲取全部論文+源代碼合集
數(shù)據(jù)（12篇）
Towards better understanding and better generalization of low-shot classification in histology images with contrastive learning
標(biāo)題：通過(guò)對(duì)比學(xué)習(xí)來(lái)更好地理解和提高組織病理圖像中的小樣本分類(lèi)的泛化能力
方法介紹：本文通過(guò)設(shè)置三個(gè)跨域任務(wù)來(lái)推動(dòng)組織病理圖像的小樣本學(xué)習(xí)研究，模擬實(shí)際臨床問(wèn)題。為實(shí)現(xiàn)高效標(biāo)注和更好的泛化能力，作者提出結(jié)合對(duì)比學(xué)習(xí)和潛在增強(qiáng)來(lái)構(gòu)建小樣本系統(tǒng)。對(duì)比學(xué)習(xí)可以在無(wú)手動(dòng)標(biāo)注下學(xué)習(xí)有用表示，而潛在增強(qiáng)以非監(jiān)督方式傳遞基數(shù)據(jù)集的語(yǔ)義變化。這兩者可充分利用無(wú)標(biāo)注訓(xùn)練數(shù)據(jù)，可擴(kuò)展到其他數(shù)據(jù)饑渴問(wèn)題。
FlipDA: Effective and robust data augmentation for few-shot learning
PromDA: Prompt-based data augmentation for low-resource NLU tasks
Generating representative samples for few-shot classification
FeLMi : Few shot learning with hard mixup
Understanding cross-domain few-shot learning based on domain similarity and few-shot difficulty
Label hallucination for few-shot classification
STUNT: Few-shot tabular learning with self-generated tasks from unlabeled tables
Unsupervised meta-learning via few-shot pseudo-supervised contrastive learning
Progressive mix-up for few-shot supervised multi-source domain transfer
Cross-level distillation and feature denoising for cross-domain few-shot classification
Tuning language models as training data generators for augmentation-enhanced few-shot learning
模型（35篇）
多任務(wù)學(xué)習(xí)
When does self-supervision improve few-shot learning?
標(biāo)題：自監(jiān)督學(xué)習(xí)在什么情況下可以改進(jìn)小樣本學(xué)習(xí)？
方法介紹：雖然自監(jiān)督學(xué)習(xí)的收益可能隨著更大的訓(xùn)練數(shù)據(jù)集而增加，但我們也觀察到，當(dāng)用于元學(xué)習(xí)和自監(jiān)督的圖像分布不同時(shí)，自監(jiān)督學(xué)習(xí)實(shí)際上可能會(huì)損害性能。通過(guò)系統(tǒng)地變化域移度和在多個(gè)域上分析幾種元學(xué)習(xí)算法的性能，作者進(jìn)行了詳細(xì)的分析研究?；谶@一分析，作者提出了一種從大規(guī)模通用無(wú)標(biāo)注圖像池中自動(dòng)選擇適合特定數(shù)據(jù)集的自監(jiān)督學(xué)習(xí)圖像的技術(shù)，可以進(jìn)一步改進(jìn)性能。
Pareto self-supervised training for few-shot learning
Bridging multi-task learning and meta-learning: Towards efficient training and effective adaptation
Task-level self-supervision for cross-domain few-shot learning
嵌入/度量學(xué)習(xí)
Few-shot learning as cluster-induced voronoi diagrams: A geometric approach
標(biāo)題：將小樣本學(xué)習(xí)視為由簇誘導(dǎo)的Voronoi圖：一種幾何方法
方法介紹：小樣本學(xué)習(xí)仍面臨泛化能力不足的挑戰(zhàn)，本文從幾何視角出發(fā)，發(fā)現(xiàn)流行的 ProtoNet 模型本質(zhì)上是特征空間中的 Voronoi 圖。通過(guò)利用“由簇誘導(dǎo)的 Voronoi 圖”的技術(shù)，可以逐步改進(jìn)空間分割，在小樣本學(xué)習(xí)的多個(gè)階段提升準(zhǔn)確率和魯棒性。這一基于該圖的框架數(shù)學(xué)優(yōu)雅、幾何可解釋，可以補(bǔ)償極端數(shù)據(jù)不足，防止過(guò)擬合，并實(shí)現(xiàn)快速幾何推理。
Few-shot learning with siamese networks and label tuning
Matching feature sets for few-shot image classification
EASE: Unsupervised discriminant subspace learning for transductive few-shot learning
Cross-domain few-shot learning with task-specific adapters
Rethinking generalization in few-shot classification
Hybrid graph neural networks for few-shot learning
Hubs and hyperspheres: Reducing hubness and improving transductive few-shot learning with hyperspherical embeddings
Revisiting prototypical network for cross domain few-shot learning
Transductive few-shot learning with prototype-based label propagation by iterative graph refinement
Few-sample feature selection via feature manifold learning
Interval bound interpolation for few-shot learning with few tasks
A closer look at few-shot classification again
TART: Improved few-shot text classification using task-adaptive reference transformation
外部存儲(chǔ)器輔助學(xué)習(xí)
Dynamic memory induction networks for few-shot text classification
標(biāo)題：動(dòng)態(tài)記憶誘導(dǎo)網(wǎng)絡(luò)用于短文本分類(lèi)
方法介紹：本文提出了動(dòng)態(tài)記憶誘導(dǎo)網(wǎng)絡(luò)，用于短文本的少樣本分類(lèi)。該模型利用動(dòng)態(tài)路由為基于記憶的少樣本學(xué)習(xí)提供更大靈活性，以便更好地適應(yīng)支持集，這是少樣本分類(lèi)模型的關(guān)鍵能力。在此基礎(chǔ)上，作者進(jìn)一步開(kāi)發(fā)了包含查詢信息的誘導(dǎo)模型，旨在增強(qiáng)元學(xué)習(xí)的泛化能力。
Few-shot visual learning with contextual memory and fine-grained calibration
Learn from concepts: Towards the purified memory for few-shot learning
Prototype memory and attention mechanisms for few shot image generation
Hierarchical variational memory for few-shot learning across domains
Remember the difference: Cross-domain few-shot semantic segmentation via meta-memory transfer
Consistent prototype learning for few-shot continual relation extraction
生成式建模
Few-shot relation extraction via bayesian meta-learning on relation graphs
標(biāo)題：通過(guò)關(guān)系圖上的貝葉斯元學(xué)習(xí)實(shí)現(xiàn)短文本關(guān)系提取
方法介紹：作者提出了一種新的貝葉斯元學(xué)習(xí)方法，用于有效學(xué)習(xí)關(guān)系原型向量的后驗(yàn)分布，其中關(guān)系原型向量的先驗(yàn)由定義在全局關(guān)系圖上的圖神經(jīng)網(wǎng)絡(luò)參數(shù)化。此外，為了有效優(yōu)化原型向量的后驗(yàn)分布，作者使用了相關(guān)于MAML算法的隨機(jī)梯度蘭weibo乎動(dòng)力學(xué)，它可以處理原型向量的不確定性，整個(gè)框架可以端到端高效優(yōu)化。
Interventional few-shot learning
Modeling the probabilistic distribution of unlabeled data for one-shot medical image segmentation
SCHA-VAE: Hierarchical context aggregation for few-shot generation
Diversity vs. Recognizability: Human-like generalization in one-shot generative models
Generalized one-shot domain adaptation of generative adversarial networks
Towards diverse and faithful one-shot adaption of generative adversarial networks
Few-shot cross-domain image generation via inference-time latent-code learning
Adaptive IMLE for few-shot pretraining-free generative modelling
MetaModulation: Learning variational feature hierarchies for few-shot learning with fewer tasks
算法（24篇）
優(yōu)化已有參數(shù)
Revisit finetuning strategy for few-shot learning to transfer the emdeddings
Prototypical calibration for few-shot learning of language models
Prompt, generate, then cache: Cascade of foundation models makes strong few-shot learners
Supervised masked knowledge distillation for few-shot transformers
Hint-Aug: Drawing hints from foundation vision transformers towards boosted few-shot parameter-efficient tuning
Few-shot learning with visual distribution calibration and cross-modal distribution alignment
MetricPrompt: Prompting model as a relevance metric for few-shot text classification
Multitask pre-training of modular prompt for chinese few-shot learning
Cold-start data selection for better few-shot language model fine-tuning: A prompt-based uncertainty propagation approach
Instruction induction: From few examples to natural language task descriptions
Hierarchical verbalizer for few-shot hierarchical text classification
優(yōu)化元學(xué)習(xí)中的參數(shù)
How to train your MAML to excel in few-shot classification
Meta-learning with fewer tasks through task interpolation
Dynamic kernel selection for improved generalization and memory efficiency in meta-learning
What matters for meta-learning vision regression tasks?
Stochastic deep networks with linear competing units for model-agnostic meta-learning
Robust meta-learning with sampling noise and label noise via Eigen-Reptile
Attentional meta-learners for few-shot polythetic classification
PLATINUM: Semi-supervised model agnostic meta-learning using submodular mutual information
FAITH: Few-shot graph classification with hierarchical task graphs
A contrastive rule for meta-learning
Meta-ticket: Finding optimal subnetworks for few-shot learning within randomly initialized neural networks
學(xué)習(xí)搜索步驟
Optimization as a model for few-shot learning
Meta Navigator: Search for a good adaptation policy for few-shot learning
應(yīng)用（57篇）
計(jì)算機(jī)視覺(jué)
Analogy-forming transformers for few-shot 3D parsing
Universal few-shot learning of dense prediction tasks with visual token matching
Meta learning to bridge vision and language models for multimodal few-shot learning
Few-shot geometry-aware keypoint localization
AsyFOD: An asymmetric adaptation paradigm for few-shot domain adaptive object detection
A strong baseline for generalized few-shot semantic segmentation
StyleAdv: Meta style adversarial training for cross-domain few-shot learning
DiGeo: Discriminative geometry-aware learning for generalized few-shot object detection
Hierarchical dense correlation distillation for few-shot segmentation
CF-Font: Content fusion for few-shot font generation
MoLo: Motion-augmented long-short contrastive learning for few-shot action recognition
MIANet: Aggregating unbiased instance and general information for few-shot semantic segmentation
FreeNeRF: Improving few-shot neural rendering with free frequency regularization
Exploring incompatible knowledge transfer in few-shot image generation
Where is my spot? few-shot image generation via latent subspace optimization
FGNet: Towards filling the intra-class and inter-class gaps for few-shot segmentation
GeCoNeRF: Few-shot neural radiance fields via geometric consistency
掃碼添加小享，回復(fù)“小樣本160”
免費(fèi)獲取全部論文+源代碼合集
機(jī)器人技術(shù)
One solution is not all you need: Few-shot extrapolation via structured MaxEnt RL
Bowtie networks: Generative modeling for joint few-shot recognition and novel-view synthesis
Demonstration-conditioned reinforcement learning for few-shot imitation
Hierarchical few-shot imitation with skill transition models
Prompting decision transformer for few-shot policy generalization
Stage conscious attention network (SCAN): A demonstration-conditioned policy for few-shot imitation
Online prototype alignment for few-shot policy transfer
自然語(yǔ)言處理
A dual prompt learning framework for few-shot dialogue state tracking
CLUR: Uncertainty estimation for few-shot text classification with contrastive learning
Few-shot document-level event argument extraction
MetaAdapt: Domain adaptive few-shot misinformation detection via meta learning
Code4Struct: Code generation for few-shot event structure prediction
MANNER: A variational memory-augmented model for cross domain few-shot named entity recognition
Few-shot event detection: An empirical study and a unified view
CodeIE: Large code generation models are better few-shot information extractors
Few-shot in-context learning on knowledge base question answering
Linguistic representations for fewer-shot relation extraction across domains
Few-shot reranking for multi-hop QA via language model prompting
知識(shí)圖譜
Adaptive attentional network for few-shot knowledge graph completion
Learning inter-entity-interaction for few-shot knowledge graph completion
Few-shot relational reasoning via connection subgraph pretraining
Hierarchical relational learning for few-shot knowledge graph completion
The unreasonable effectiveness of few-shot learning for machine translation
聲音信號(hào)處理
Audio2Head: Audio-driven one-shot talking-head generation with natural head motion
Few-shot low-resource knowledge graph completion with multi-view task representation generation
Normalizing flow-based neural process for few-shot knowledge graph completion
推薦系統(tǒng)
Few-shot news recommendation via cross-lingual transfer
ColdNAS: Search to modulate for user cold-start recommendation
Contrastive collaborative filtering for cold-start item recommendation
SMINet: State-aware multi-aspect interests representation network for cold-start users recommendation
Multimodality helps unimodality: Cross-modal few-shot learning with multimodal models
M2EU: Meta learning for cold-start recommendation via enhancing user preference estimation
Aligning distillation for cold-start item recommendation
其他
Context-enriched molecule representations improve few-shot drug discovery
Sequential latent variable models for few-shot high-dimensional time-series forecasting
Transfer NAS with meta-learned Bayesian surrogates
Few-shot domain adaptation for end-to-end communication
Contrastive meta-learning for few-shot node classification
Task-equivariant graph few-shot learning
Leveraging transferable knowledge concept graph embedding for cold-start cognitive diagnosis
理論（7篇）
Bridging the gap between practice and PAC-Bayes theory in few-shot meta-learning
bounds for meta-learning: An information-theoretic analysis
Generalization bounds for meta-learning via PAC-Bayes and uniform stability
Unraveling model-agnostic meta-learning via the adaptation learning rate
On the importance of firth bias reduction in few-shot classification
Global convergence of MAML and theory-inspired neural architecture search for few-shot learning
Smoothed embeddings for certified few-shot learning
小樣本/零樣本學(xué)習(xí)（15篇）
Finetuned language models are zero-shot learners
Zero-shot stance detection via contrastive learning
JointCL: A joint contrastive learning framework for zero-shot stance detection
Knowledgeable prompt-tuning: Incorporating knowledge into prompt verbalizer for text classification
Nearest neighbor zero-shot inference
Continued pretraining for better zero- and few-shot promptability
InstructDial: Improving zero and few-shot generalization in dialogue through instruction tuning
Prompt-and-Rerank: A method for zero-shot and few-shot arbitrary textual style transfer with small language models
Learning instructions with unlabeled data for zero-shot cross-task generalization
Zero-shot cross-lingual transfer of prompt-based tuning with a unified multilingual prompt
Finetune like you pretrain: Improved finetuning of zero-shot vision models
SemSup-XC: Semantic supervision for zero and few-shot extreme classification
Zero- and few-shot event detection via prompt-based meta learning
HINT: Hypernetwork instruction tuning for efficient zero- and few-shot generalisation
What does the failure to reason with "respectively" in zero/few-shot settings tell us about language models? acl 2023
小樣本學(xué)習(xí)變體（12篇）
FiT: Parameter efficient few-shot transfer learning for personalized and federated image classification
Towards addressing label skews in one-shot federated learning
Data-free one-shot federated learning under very high statistical heterogeneity
Contrastive meta-learning for partially observable few-shot learning
On the soft-subnetwork for few-shot class incremental learning
Warping the space: Weight space rotation for class-incremental few-shot learning
Neural collapse inspired feature-classifier alignment for few-shot class-incremental learning
Learning with fantasy: Semantic-aware virtual contrastive constraint for few-shot class-incremental learning
Few-shot class-incremental learning via class-aware bilateral distillation
Glocal energy-based learning for few-shot open-set recognition
Open-set likelihood maximization for few-shot learning
Federated few-shot learning
數(shù)據(jù)集/基準(zhǔn)（5篇）
FewNLU: Benchmarking state-of-the-art methods for few-shot natural language understanding
Bongard-HOI: Benchmarking few-shot visual reasoning for human-object interactions
Hard-Meta-Dataset++: Towards understanding few-shot performance on difficult tasks
MEWL: Few-shot multimodal word learning with referential uncertainty
UNISUMM and SUMMZOO: Unified model and diverse benchmark for few-shot summarization
掃碼添加小享，回復(fù)“小樣本160”
免費(fèi)獲取全部論文+源代碼合集
