遷移學習相關資料

轉自王晉東的部落格

遷移學習簡介

遷移學習(transfer learning)通俗來講，就是運用已有的知識來學習新的知識，核心是找到已有知識和新知識之間的相似性，用成語來說就是舉一反三。由于直接對目标域從頭開始學習成本太高，我們故而轉向運用已有的相關知識來輔助盡快地學習新知識。比如，已經會下中國象棋，就可以類比着來學習國際象棋；已經會編寫Java程式，就可以類比着來學習C#；已經學會英語，就可以類比着來學習法語；等等。世間萬事萬物皆有共性，如何合理地找尋它們之間的相似性，進而利用這個橋梁來幫助學習新知識，是遷移學習的核心問題。

圖1不同位置、不同傳感器的遷移标定。已知一個房間中A點的WiFi信号與相應的人體行為，如何标定另一個房間中C點的藍牙信号？

具體地，在遷移學習中，我們已有的知識叫做源域(source domain)，要學習的新知識叫目标域(target domain)。遷移學習研究如何把源域的知識遷移到目标域上。特别地，在機器學習領域中，遷移學習研究如何将已有模型應用到新的不同的、但是有一定關聯的領域中。傳統機器學習在應對資料的分布、次元，以及模型的輸出變化等任務時，模型不夠靈活、結果不夠好，而遷移學習放松了這些假設。在資料分布、特征次元以及模型輸出變化條件下，有機地利用源域中的知識來對目标域更好地模組化。另外，在有标定資料缺乏的情況下，遷移學習可以很好地利用相關領域有标定的資料完成資料的标定。

圖2 遷移學習與傳統機器學習的不同。(a)傳統機器學習對不同的學習任務建立不同的模型，(b)遷移學習利用源域中的資料将知識遷移到目标域，完成模型建立。插圖來自：Sinno Jialin Pan and Qiang Yang, A survey on transfer learning. IEEE TKDE 2010.

遷移學習按照學習方式可以分為基于樣本的遷移，基于特征的遷移，基于模型的遷移，以及基于關系的遷移。基于樣本的遷移通過對源域中有标定樣本的權重利用完成知識遷移；基于特征的遷移通過将源域和目标域映射到相同的空間（或者将其中之一映射到另一個的空間中）并最小化源域和目标域的距離來完成知識遷移；基于模型的遷移将源域和目标域的模型與樣本結合起來調整模型的參數；基于關系的遷移則通過在源域中學習概念之間的關系，然後将其類比到目标域中，完成知識的遷移。

理論上，任何領域之間都可以做遷移學習。但是，如果源域和目标域之間相似度不夠，遷移結果并不會理想，出現所謂的負遷移情況。比如，一個人會騎自行車，就可以類比學電動車；但是如果類比着學開汽車，那就有點天方夜譚了。如何找到相似度盡可能高的源域和目标域，是整個遷移過程最重要的前提。

遷移學習方面，代表人物有香港科技大學的Qiang Yang教授，南洋理工大學的Sinno Jialin Pan，以及第四範式的CEO戴文淵等。代表文獻是Sinno Jialin Pan和Qiang Yang的A survey on transfer learning。

作者網站：http://jd92.wang.

[參考資料]

[1] Pan S J, Yang Q. A survey on transfer learning[J]. IEEE Transactions on knowledge and data engineering, 2010, 22(10): 1345-1359.

[2] Introduction to Transfer Learning: http://jd92.wang/assets/files/l03_transferlearning.pdf。

[3] Qiang Yang: http://www.cse.ust.hk/~qyang/.

[4] Sinno Jialin Pan: http://www.ntu.edu.sg/home/sinnopan/.

[5] Wenyuan Dai: https://scholar.google.com/citations?user=AGR9pP0AAAAJ&hl=zh-CN.

Transfer learning applications

遷移學習的應用

• 20191222 NIPS-19 workshop Sim-to-Real Domain Adaptation For High Energy Physics
  • Transfer learning for high energy physics
  • 遷移學習用于高能實體
• 20191222 arXiv Transfer learning in hybrid classical-quantum neural networks
  • Transfer learning for quantum neural networks
• 20191214 arXiv Unsupervised Transfer Learning via BERT Neuron Selection
  • Unsupervised transfer learning via BERT neuron selection
• 20191214 arXiv Transfer Learning-Based Outdoor Position Recovery with Telco Data
  • Outdoor position recorvey with Telco data using transfer learning
• 20191214 NIPS-19 workshop Cross-Language Aphasia Detection using Optimal Transport Domain Adaptation
  • Optimal transport domain adaptation
• 20191201 arXiv A Transfer Learning Method for Goal Recognition Exploiting Cross-Domain Spatial Features
  • A transfer learning method for goal recognition
  • 用遷移學習分析語言中的目标
• 20191201 AAAI-20 Zero-Resource Cross-Lingual Named Entity Recognition
  • Zero-resource cross-lingual NER
  • 零資源的跨語言NER
• 20191125 arXiv Attention Privileged Reinforcement Learning For Domain Transfer
  • Attention privileged reinforcement learning for domain transfer
• 20191124 Cantonese Automatic Speech Recognition Using Transfer Learning from Mandarin
  • Cantonese speech recognition using transfer learning from mandarin
  • 國語語音識别遷移到廣東話識别
• 20191115 arXiv Instance-based Transfer Learning for Multilingual Deep Retrieval
  • Instance based transfer learning for multilingual deep retrieval
  • 基于執行個體的遷移學習用于多語言的retrieval
• 20191115 arXiv Unsupervised Pre-training for Natural Language Generation: A Literature Review
  • Unsupervised pre-training for natural language generation survey
  • 一篇無監督預訓練用于自然語言生成的綜述
• 20191115 AAAI-20 Unsupervised Domain Adaptation on Reading Comprehension
  • 無監督DA用于閱讀了解
  • Unsupervised DA for reading comprehension
• 20191113 arXiv Open-Ended Visual Question Answering by Multi-Modal Domain Adaptation
  • Supervised multi-modal domain adaptation in VQA
  • 有監督的多模态DA用于VQA任務
• 20191113 arXiv NegBERT: A Transfer Learning Approach for Negation Detection and Scope Resolution
  • Transfer learning for negation detection and scope resolution
  • 遷移學習用于否定檢測
• 20191113 AAAI-20 TANDA: Transfer and Adapt Pre-Trained Transformer Models for Answer Sentence Selection
  • Finetune twice for answer sentence selection
  • 兩次finetune用于answer sentence selection
• 20191111 NIPS-19 workshop Transfer Learning in 4D for Breast Cancer Diagnosis using Dynamic Contrast-Enhanced Magnetic Resonance Imaging
  • Transfer learning in 4D for breast cancer diagnosis
• 20191111 BigData-19 Deep Transfer Learning for Thermal Dynamics Modeling in Smart Buildings
  • Transfer learning for thermal dynamics modeling
• 20191111 arXiv Unsupervised Domain Adaptation of Contextual Embeddings for Low-Resource Duplicate Question Detection
  • Unsupervised DA for low-resource duplicate question detection
• 20191111 arXiv Towards Domain Adaptation from Limited Data for Question Answering Using Deep Neural Networks
  • DA for question answering using DNN
• 20191111 arXiv Teacher-Student Training for Robust Tacotron-based TTS
  • Teacher-student network for robust TTS
• 20191111 arXiv SMART: Robust and Efficient Fine-Tuning for Pre-trained Natural Language Models through Principled Regularized Optimization
  • Fine-tuning for pre-trained language model
• 20191111 arXiv Change your singer: a transfer learning generative adversarial framework for song to song conversion
  • Adversarial transfer learning for song-to-song conversion
• 20191111 arXiv Transfer Learning in Spatial-Temporal Forecasting of the Solar Magnetic Field
  • Transfer learning for solar magnetic field
• 20191111 arXiv Deep geometric knowledge distillation with graphs
  • Deep geometric knowledge distillation with graphs
• 20191101 arXiv Transferable End-to-End Aspect-based Sentiment Analysis with Selective Adversarial Learning
  • Adversarial transfer learning for aspect-based sentement analysis
  • 對抗遷移用于aspect層級的情感分析
• 20191101 Transfer Learning from Transformers to Fake News Challenge Stance Detection (FNC-1) Task
  • A fake news challenges based on transformers
  • 一個基于transformer的假新聞檢測挑戰
• 20191029 arXiv NER Models Using Pre-training and Transfer Learning for Healthcare
  • Pretraining NER models for healthcare
  • 預訓練的NER模型用于健康監護
• 20191029 WACV-20 Progressive Domain Adaptation for Object Detection
  • Progressive domain adaptation for object recognition
  • 漸進式的DA用于物體檢測
• 20191029 WSDM-20 Meta-Learning with Dynamic-Memory-Based Prototypical Network for Few-Shot Event Detection
  • Meta learning with dynamic memory based prototypical network for few-shot event detection
• 20191017 arXiv Evolution of transfer learning in natural language processing
  • Survey transfer learning works in NLP
  • 綜述了最近遷移學習在NLP的一些進展
• 20191017 arXiv Unsupervised Domain Adaptation Meets Offline Recommender Learning
  • Unsupervised DA meets offline recommender learning
  • 無監督DA用于離線推薦系統
• 20191017 Transfer Learning for Algorithm Recommendation
  • Transfer learning for algorithm recommendation
  • 遷移學習用于算法推薦
• 20191015 arXiv Emotion Recognition in Conversations with Transfer Learning from Generative Conversation Modeling
  • Emotion recognition in conversations with transfer learning
  • 用遷移學習進行對話中的情緒識别
• 20191015 WSDM-20 DDTCDR: Deep Dual Transfer Cross Domain Recommendation
  • Cross-modal recommendation using dual transfer learning
  • 用對偶遷移進行跨模态推薦
• 20191011 NeurIPS-19 Unified Language Model Pre-training for Natural Language Understanding and Generation
  • Unified language model pre-training for understanding and generation
  • 統一的語言模型預訓練用于自然語言了解和生成
• 20191011 ICIP-19 Cross-modal knowledge distillation for action recognition
  • Cross-modal knowledge distillation for action recognition
  • 跨模态的知識蒸餾并用于動作識别
• 20191011 NeurIPS-19 workshop Language Transfer for Early Warning of Epidemics from Social Media
  • Language transfer to predict epidemics from social media
  • 通過社交網絡資料預測傳染病并進行語言模型的遷移
• 20191008 arXiv, ICCV-19 demo Cross-Domain Complementary Learning with Synthetic Data for Multi-Person Part Segmentation
  • Learning human body part segmentation without human labeling
  • 基於合成數據的跨域互補學習人體部位分割
• 20191008 arXiv Transfer Brain MRI Tumor Segmentation Models Across Modalities with Adversarial Networks
  • Transfer learning for multi-modal brain MRI tumor segmentation
  • 用遷移學習進行多模态的MRI惡性良性腫瘤分割
• 20191008 ICONIP-19 Semi-Supervised Domain Adaptation with Representation Learning for Semantic Segmentation across Time
  • Semi-supervised domain adaptation with representation learning for semantic segmentation
  • 半監督DA用于語義分割
• 20191008 arXiv Noise as Domain Shift: Denoising Medical Images by Unpaired Image Translation
  • Noise as domain shift for medical images
  • 醫學圖像中的噪聲進行adaptation
• 20190926 arXiv Restyling Data: Application to Unsupervised Domain Adaptation
  • Restyle data using domain adaptation
  • 使用domain adaptation進行風格遷移
• 20190916 ISWC-19 Cross-dataset deep transfer learning for activity recognition
  • Cross-dataset transfer learning for activity recognition
  • 跨資料集的深度遷移學習用于行為識别
• 20190912 MICCAI workshop Multi-Domain Adaptation in Brain MRI through Paired Consistency and Adversarial Learning
  • Multi-domain adaptation for brain MRI
  • 多領域的adaptation用于大腦MRI識别
• 20190909 IJCAI-FML-19 FedHealth: A Federated Transfer Learning Framework for Wearable Healthcare
  • The first work on federated transfer learning for wearable healthcare
  • 第一個将聯邦遷移學習用于可穿戴健康監護的工作
• 20190829 EMNLP-19 Investigating Meta-Learning Algorithms for Low-Resource Natural Language Understanding Tasks
  • Investigating MAML for low-resource NMT
  • 調查了MAML方法用于低資源的NMT問題的表現
• 20190829 EMNLP-19 Unsupervised Domain Adaptation for Neural Machine Translation with Domain-Aware Feature Embeddings
  • Domain adaptation for NMT
• 20190828 MICCAI-19 workshop Cross-modality Knowledge Transfer for Prostate Segmentation from CT Scans
  • Cross-modality transfer for prostate segmentation
  • 跨模态的遷移用于前列腺分割
• 20180828 ICCV-19 workshop Unsupervised Deep Feature Transfer for Low Resolution Image Classification
  • Deep feature transfer for low resolution image classification
  • 深度特征遷移用于低分辨率圖像分類
• 20190828 arXiv VAE-based Domain Adaptation for Speaker Verification
  • Speaker verification using VAE domain adaptation
  • 基于VAE的speaker verification
• 20190821 arXiv Shallow Domain Adaptive Embeddings for Sentiment Analysis
  • Domain adaptative embedding for sentiment analysis
  • 遷移學習用于情感分類
• 20190813 IJAIT Transferring knowledge from monitored to unmonitored areas for forecasting parking spaces
  • Transfer learning for forecasting parking spaces
  • 用遷移學習預測停車空間
• 20190809 ICCASP-19 Cross-lingual Text-independent Speaker Verification using Unsupervised Adversarial Discriminative Domain Adaptation
  • Text independent speaker verification using adversarial DA
  • 文本無關的speaker verification用DA
• 20190809 IJCAI-19 Progressive Transfer Learning for Person Re-identification
  • Progressive transfer learning for RE_ID
  • 漸進式遷移學習用于RE_ID
• 20190809 NeurIPS-18 MacNet: Transferring Knowledge from Machine Comprehension to Sequence-to-Sequence Models
  • Transfer learning from machine comprehension to sequence to senquence Models
  • 從機器了解到序列模型遷移
• 20190802 arXiv Towards More Accurate Automatic Sleep Staging via Deep Transfer Learning
  • Accurate Sleep Staging with deep transfer learning
  • 用深度遷移學習進行精準的睡眠階段估計
• 20190729 MICCAI-19 Annotation-Free Cardiac Vessel Segmentation via Knowledge Transfer from Retinal Images
  • Cardiac vessel segmentation using transfer learning from Retinal Images
  • 用視網膜圖檔進行遷移學習用于心髒血管分割
• 20190703 arXiv Disentangled Makeup Transfer with Generative Adversarial Network
  • Makeup transfer with GAN
  • 用GAN進行化妝的遷移
• 20190703 arXiv Applying Transfer Learning To Deep Learned Models For EEG Analysis
  • Apply transfer learning to EEG
  • 用深度遷移學習進行EEG分析
• 20190626 arXiv A Novel Deep Transfer Learning Method for Detection of Myocardial Infarction
  • A deep transfer learning method for detecting myocardial infarction
  • 一種用于監測心肌梗塞的深度遷移方法
• 20190517 PHM-19 Domain Adaptive Transfer Learning for Fault Diagnosis
  • Domain adaptation for fault diagnosis
  • 領域自适應用于錯誤檢測
• 20190515 ACL-19 Effective Cross-lingual Transfer of Neural Machine Translation Models without Shared Vocabularies
  • Cross-lingual transfer of NMT
  • 跨語言的NMT模型遷移
• 20190509 arXiv Unsupervised Domain Adaptation using Generative Adversarial Networks for Semantic Segmentation of Aerial Images
  • Domain adaptation for semantic segmentation in aerial images
  • DA應用于鳥瞰圖像語義分割
• 20190508 arXiv Text2Node: a Cross-Domain System for Mapping Arbitrary Phrases to a Taxonomy
  • Cross-domain system for mapping arbitrary phrases to a taxonomy
• 20190508 arXiv On Transfer Learning For Chatter Detection in Turning Using Wavelet Packet Transform and Empirical Mode Decomposition
  • Transfer learning for chatter detection
  • 用遷移學習進行叽叽喳喳聊天識别
• 20190416 arXiv Deep Transfer Learning for Single-Channel Automatic Sleep Staging with Channel Mismatch
  • Using deep transfer learning for sleep stage recognition
  • 用深度遷移學習進行睡眠階段的檢測
• 20190415 PAKDD-19 Targeted Knowledge Transfer for Learning Traffic Signal Plans
  • Targeted knowledge transfer for traffic control
  • 目标知識遷移應用于交通紅綠燈
• 20190415 PAKDD-19 Knowledge Graph Rule Mining via Transfer Learning
  • Knowledge Graph Rule Mining via Transfer Learning
  • 遷移學習應用于知識圖譜
• 20190415 PAKDD-19 Adaptively Transfer Category-Classifier for Handwritten Chinese Character Recognition
  • Transfer learning for handwritten Chinese character recognition
  • 用遷移學習進行中文手寫體識别
• 20190415 PAKDD-19 Multi-task Learning for Target-Dependent Sentiment Classification
  • Multi-task learning for sentiment classification
  • 用多任務學習進行任務依賴的情感分析
• 20190415 PAKDD-19 Spatial-Temporal Multi-Task Learning for Within-Field Cotton Yield Prediction
  • Spatial-Temporal multi-task learning for cotton yield prediction
  • 時空依賴的多任務學習用于棉花收入預測
• 20190415 PAKDD-19 Passenger Demand Forecasting with Multi-Task Convolutional Recurrent Neural Networks
  • Passenger demand forecasting with multi-task CRNN
  • 用多任務CRNN模型進行顧客需求估計
• 20190409 arXiv Unsupervised Domain Adaptation for Multispectral Pedestrian Detection
  • Domain adaptation for pedestrian detection
  • 無監督領域自适應用于多模态行人檢測
• 20190408 arXiv Unsupervised Domain Adaptation of Contextualized Embeddings: A Case Study in Early Modern English
  • Domain adaptation in early modern english
  • Case study: 在英文中的domain adaptation
• 20190408 USENIX-19 Transfer Learning for Performance Modeling of Deep Neural Network Systems
  • Using transfer learning for performance modeling in DNN
  • 在深度網絡中使用遷移學習進行性能模組化
• 20190403 arXiv Transfer Learning for Clinical Time Series Analysis using Deep Neural Networks
  • Using transfer learning for multivariate clinical data
  • 使用遷移學習進行多元醫療資料遷移
• 20190403 arXiv Med3D: Transfer Learning for 3D Medical Image Analysis
  • Transfer learning for 3D medical image analysis
  • 遷移學習用于3D醫療圖像分析
• 20190401 arXiv Cross-Subject Transfer Learning in Human Activity Recognition Systems using Generative Adversarial Networks
  • Cross-subject transfer learning using GAN
  • 用對抗網絡進行跨使用者的行為識别
• 20190305 arXiv Unsupervised Domain Adaptation Learning Algorithm for RGB-D Staircase Recognition
  • Domain adaptation for RGB-D staircase recognition
  • Domain adaptation進行深度和RGB樓梯識别
• 20190221 arXiv Transfusion: Understanding Transfer Learning with Applications to Medical Imaging
  • Analyzing the influence of transfer learning in medical imaging
  • 在醫療圖像中分析遷移學習作用
• 20190123 arXiv Transfer Learning and Meta Classification Based Deep Churn Prediction System for Telecom Industry
  • Transfer learning in telcom industry
  • 遷移學習用于電信行業
• 20190123 arXiv Cold-start Playlist Recommendation with Multitask Learning
  • Cold-start playlist recommendation with multitask learning
  • 用多任務學習進行冷啟動狀态下的播放清單推薦
• 20190123 arXiv Adapting Convolutional Neural Networks for Geographical Domain Shift
  • Convolutional neural network for geographical domain shift
  • 将卷積網絡用于地理學上的domain shift問題
• 20190117 NeurIPS-18 workshop Transfer Learning for Prosthetics Using Imitation Learning
  • Using transfer learning for prosthetics
  • 用遷移學習進行義肢的模仿學習
• 20190115 IJAERS Weightless Neural Network with Transfer Learning to Detect Distress in Asphalt
  • Transfer learning to detect distress in asphalt
  • 用遷移學習檢測路面情況
• 20190115 arXiv Disease Knowledge Transfer across Neurodegenerative Diseases
  • Transfer learning for neurodegenerative disease
  • 遷移學習用于神經退行性疾病
• 20190111 ICMLA-18 Supervised Transfer Learning for Product Information Question Answering
  • Transfer learning for product information question answering
  • 利用遷移學習進行産品資訊的對話
• 20190102 arXiv High Quality Monocular Depth Estimation via Transfer Learning
  • Monocular depth estimation using transfer learning
  • 用遷移學習進行單眼深度估計
• 20181230 arXiv The CORAL+ Algorithm for Unsupervised Domain Adaptation of PLDA
  • Use CORAL for speaker recognition
  • 用CORAL改進版進行speaker識别
• 20181230 arXiv Domain-Aware Generalized Zero-Shot Learning
  • Domain-aware zero-shot learning
• 20181225 arXiv A Multi-task Neural Approach for Emotion Attribution, Classification and Summarization
  • A multi-task approach for emotion attribution, classification, and summarization
  • 一個多任務方法同時用于情緒歸屬、分類和總結
• 20181225 arXiv A General Approach to Domain Adaptation with Applications in Astronomy
  • Adopting active learning to transfer model
  • 用主動學習來進行模型遷移并應用到天文學上
• 20181225 arXiv An Integrated Transfer Learning and Multitask Learning Approach for Pharmacokinetic Parameter Prediction
  • Using transfer learning for Pharmacokinetic Parameter Prediction
  • 用遷移學習進行藥代動力學參數估計
• 20181221 arXiv Deep Transfer Learning for Static Malware Classification
  • Deep Transfer Learning for Static Malware Classification
  • 用深度遷移學習進行惡意軟體分類
• 20181221 arXiv PnP-AdaNet: Plug-and-Play Adversarial Domain Adaptation Network with a Benchmark at Cross-modality Cardiac Segmentation
  • Adversarial transfer learning for medical images
  • 對抗遷移學習用于醫學圖像分割
• 20181220 arXiv Domain Adaptation for Reinforcement Learning on the Atari
  • Reinforcement domain adaptation on Atari games
  • 遷移強化學習用于Atari遊戲
• 20181220 arXiv Deep UL2DL: Channel Knowledge Transfer from Uplink to Downlink
  • Channel knowledge transfer in CSI
  • Wifi定位中的知識遷移
• 20181219 NER-19 Transfer Learning in Brain-Computer Interfaces with Adversarial Variational Autoencoders
  • Transfer learning in brain-computer interfaces
  • 遷移學習在腦機互動中的應用
• 20181219 ICCPS-19 Simulation to scaled city: zero-shot policy transfer for traffic control via autonomous vehicles
  • Transfer learning in autonomous vehicles
  • 遷移學習用于自動駕駛車輛的政策遷移
• 20181218 arXiv Transfer learning to model inertial confinement fusion experiments
  • Using transfer learning for inertial confinement fusion
  • 用遷移學習進行慣性限制聚變
• 20181214 arXiv Bridging the Generalization Gap: Training Robust Models on Confounded Biological Data
  • Transfer learning for generalizing on biological data
  • 用遷移學習增強生物資料的泛化能力
• 20181214 BioCAS-19 ECG Arrhythmia Classification Using Transfer Learning from 2-Dimensional Deep CNN Features
  • Deep transfer learning for EEG Arrhythmia Classification
  • 深度遷移學習用于心率不齊分類
• 20181214 LAK-19 Transfer Learning using Representation Learning in Massive Online Open Courses
  • Transfer learning in MOOCs
  • 遷移學習用于大規模線上網絡課程
• 20181214 DVPBA-19 Considering Race a Problem of Transfer Learning
  • Consider race in transfer learning
  • 在遷移學習問題中考慮種族問題(跨種族遷移)
• 20181213 arXiv Multichannel Semantic Segmentation with Unsupervised Domain Adaptation
  • Robot vision semantic segmentation with domain adaptation
  • 用于機器視覺中語義分割的domain adaptation
• 20181212 arXiv 3D Scene Parsing via Class-Wise Adaptation
  • Class-wise adaptation for 3D scene parsing
  • 類别的适配用于3D場景分析
• 20181212 arXiv Secure Federated Transfer Learning
  • Federated transfer learning + Encryption
  • 聯邦遷移學習+加密(楊強團隊)
• 20181206 NeurIPS-18 workshop Towards Continuous Domain adaptation for Healthcare
  • English: Continuous domain adaptation for healthcare
  • 中文：連續的domain adaptation用于健康監護
• 20181206 NeurIPS-18 workshop A Hybrid Instance-based Transfer Learning Method
  • English: Instance-based transfer learning for healthcare
  • 中文：基于樣本的遷移學習用于健康監護
• 20181205 arXiv Learning from a tiny dataset of manual annotations: a teacher/student approach for surgical phase recognition
  • English: Transfer learning for surgical phase recognition
  • 中文：遷移學習用于外科手術階段識别
• 20181204 arXiv From Known to the Unknown: Transferring Knowledge to Answer Questions about Novel Visual and Semantic Concepts
  • English: Transfer learning for VQA
  • 中文：用遷移學習進行VQA任務
• 20181128 arXiv Cross-domain Deep Feature Combination for Bird Species Classification with Audio-visual Data
  • English: Cross-domain deep feature combination for bird species classification
  • 中文：跨領域的鳥分類
• 20181128 NeurIPS-18 workshop Multi-Task Generative Adversarial Network for Handling Imbalanced Clinical Data
  • English: Multi-task learning for imbalanced clinical data
  • 中文：多任務學習用于不平衡的就診資料
• 20181128 WACV-19 CNN based dense underwater 3D scene reconstruction by transfer learning using bubble database
  • English: Transfer learning for underwater 3D scene reconstruction
  • 中文：用遷移學習進行水下3D場景重建
• 20181127 NeurIPS-18 workshop Predicting Diabetes Disease Evolution Using Financial Records and Recurrent Neural Networks
  • English: Predicting diabetes using financial records
  • 中文：用财務記錄預測糖尿病
• 20181123 NIPS-18 workshop Population-aware Hierarchical Bayesian Domain Adaptation
  • English: Applying domain adaptation to health
  • 中文：将domain adaptation應用于健康
• 20181121 arXiv Transferrable End-to-End Learning for Protein Interface Prediction
  • English: Transfer learning for protein interface prediction
  • 中文：用遷移學習進行蛋白質接口預測
• 20181121 NSFREU-18 Transfer Learning with Deep CNNs for Gender Recognition and Age Estimation
  • English: Deep transfer learning for Gender Recognition and Age Estimation
  • 中文：用深度遷移學習進行性别識别和年齡估計
• 20181121 arXiv Distribution Discrepancy Maximization for Image Privacy Preserving
  • English: Distribution Discrepancy Maximization for Image Privacy Preserving
  • 中文：通過最大化分布差異來進行圖檔隐私保護
• 20181120 arXiv Spatial-temporal Multi-Task Learning for Within-field Cotton Yield Prediction
  • English: Multi-task learning for cotton yield prediction
  • 中文：多任務學習用于棉花産量預測
• 20181117 arXiv Unsupervised domain adaptation for medical imaging segmentation with self-ensembling
  • English: Medical imaging using transfer learning
  • 中文：使用遷移學習進行醫學圖像分割
• 20181117 arXiv Performance Estimation of Synthesis Flows cross Technologies using LSTMs and Transfer Learning
  • English: Performance Estimation of Synthesis Flows cross Technologies using LSTMs and Transfer Learning
  • 中文：利用遷移學習進行合成flow評價
• 20181117 AAAI-19 GaitSet: Regarding Gait as a Set for Cross-View Gait Recognition
  • English: Cross-view gait recognition
  • 中文：跨視圖的步态識别
• 20181115 AAAI-19 Unsupervised Transfer Learning for Spoken Language Understanding in Intelligent Agents
  • English: Transfer learning for spoken language understanding
  • 中文：無監督遷移學習用于語言了解
• 20181114 arXiv A Framework of Transfer Learning in Object Detection for Embedded Systems
  • English: Transfer learning in embedded system for object detection
  • 中文：在嵌入式系統中進行針對目标檢測的遷移學習
• 20181107 ICONIP-18 Transductive Learning with String Kernels for Cross-Domain Text Classification
  • English: String kernel for cross-domain text classification using transfer learning
  • 中文：用string kernel進行遷移學習跨領域文本分類
• 20181012 arXiv Bird Species Classification using Transfer Learning with Multistage Training
  • English: Using transfer learning for bird species classification
  • 中文：用遷移學習進行鳥類分類
• 20181012 arXiv Survival prediction using ensemble tumor segmentation and transfer learning
  • English: Predicting the survival of the tumor patient using transfer learning
  • 中文：用遷移學習估計惡性良性腫瘤病人存活時間
• 20181012 ICMLA-18 Virtual Battery Parameter Identification using Transfer Learning based Stacked Autoencoder
  • English: Using transfer learning for calculating the virtual battery in a thermostatics load
  • 中文：用遷移學習進行恒溫器的電量估計
• 20180912 PervasiveHealth-18 Transfer Learning and Data Fusion Approach to Recognize Activities of Daily Life
  • English: Transfer learning to perform activity recognition using multi-model sensors
  • 中文：用多模态傳感器進行遷移學習，用于行為識别
• 20180912 ICIP-18 Adversarial Domain Adaptation with a Domain Similarity Discriminator for Semantic Segmentation of Urban Areas
  • English: Semantic segmentation using transfer learning
  • 中文：用遷移學習進行語義分割
• 20180912 arXiv Tensor Alignment Based Domain Adaptation for Hyperspectral Image Classification
  • English: Hyperspectral image classification using domain adaptation
  • 中文：用domain adaptation進行圖像分類
• 20180909 arXiv Driving Experience Transfer Method for End-to-End Control of Self-Driving Cars
  • English: Driving experience transfer on self-driving cars
  • 中文：自動駕駛車上的駕駛經驗遷移
• 20180909 arXiv Deep Learning for Domain Adaption: Engagement Recognition
  • English: deep transfer learning for engagement recognition
  • 中文：用深度遷移學習進行人機互動中的engagement識别
• 20180904 EMBC-18 Multi-Cell Multi-Task Convolutional Neural Networks for Diabetic Retinopathy Grading Kang
  • English: Use multi-task CNN for Diabetic Retinopathy Grading Kang
  • 中文：用多任務的CNN進行糖尿病的視網膜粒度檢查
• 20180904 ICPR-18 Document Image Classification with Intra-Domain Transfer Learning and Stacked Generalization of Deep Convolutional Neural Networks
  • English: Document image classification using transfer learning
  • 中文：使用遷移學習進行文檔圖像的分類
• 20180826 ISPRS journal Deep multi-task learning for a geographically-regularized semantic segmentation of aerial images
  • English: a multi-task learning network for remote sensing
  • 中文：提出一個多任務的深度網絡用于遙感圖像檢測
• 20180823 ICPR-18 Multi-task multiple kernel machines for personalized pain recognition from functional near-infrared spectroscopy brain signals
  • English: A multi-task method to recognize pains
  • 中文：提出一個multi-task架構來檢測pain
• 20180821 arXiv Unsupervised adversarial domain adaptation for acoustic scene classification
  • English: Using transfer learning for acoustic classification
  • 中文：遷移學習用于聲音場景分類
• 20180819 arXiv Transfer Learning and Organic Computing for Autonomous Vehicles
  • English: Propose different transfer learning methods to adapt the situation of autonomous driving
  • 中文：提出一些不同的遷移學習方法應用于自動駕駛的環境适配
• 20180819 arXiv Transfer Learning for Brain-Computer Interfaces: An Euclidean Space Data Alignment Approach
  • English: Propose to align the different distributions of EEG signals using transfer learning
  • 中文：針對EEG信号不同人分布不一樣的問題提出遷移學習和資料增強的方式加以解決
• 20180801 arXiv Multimodal Deep Domain Adaptation
  • English: Use multi-modal DA in robotic vision
  • 中文：在機器人視覺中使用多模态的domain adaptation
• 20180801 arXiv Rank and Rate: Multi-task Learning for Recommender Systems
  • English: A multi-task system for recommendation
  • 中文：一個針對于推薦系統的多任務學習
• 20180801 MICCAI-18 Leveraging Unlabeled Whole-Slide-Images for Mitosis Detection
  • English: Use unlabeled images for mitosis detection
  • 中文：用未标記的圖檔進行細胞有絲分裂的檢測
• 20180801 ECCV-18 DOCK: Detecting Objects by transferring Common-sense Knowledge
  • English: A method called DOCK for object detection using transfer learning
  • 中文：提出一個叫做DOCK的方法進行基于遷移學習的目标檢測
• 20180801 ECCV-18 A Zero-Shot Framework for Sketch-based Image Retrieval
  • English: A Zero-Shot Framework for Sketch-based Image Retrieval
  • 中文：一個針對于簡筆畫圖像檢索的zero-shot架構
• 20180731 ICANN-18 Metric Embedding Autoencoders for Unsupervised Cross-Dataset Transfer Learning
  • English: Deep transfer learning for Re-ID
  • 中文：将深度遷移學習用于Re-ID
• 20180705 arXiv 将遷移學習應用于自動駕駛中的不同天氣适配：Modular Vehicle Control for Transferring Semantic Information to Unseen Weather Conditions using GANs
• 20180627 arXiv 用遷移學習進行感染預測：Domain Adaptation for Infection Prediction from Symptoms Based on Data from Different Study Designs and Contexts
• 20180627 arXiv 生成模型用于姿态遷移：Generative Models for Pose Transfer
• 20180622 arXiv 跨領域的人臉識别用于銀行認證系統：Cross-Domain Deep Face Matching for Real Banking Security Systems
• 20180621 arXiv 遷移學習用于角膜組織的分類：Transfer Learning with Human Corneal Tissues: An Analysis of Optimal Cut-Off Layer
• 20180621 arXiv 遷移學習用于強化學習中的圖像翻譯：Transfer Learning for Related Reinforcement Learning Tasks via Image-to-Image Translation
• 20180615 Interspeech-18 很全面地探索了很多類方法在語音識别上的應用：A Study of Enhancement, Augmentation, and Autoencoder Methods for Domain Adaptation in Distant Speech Recognition
• 20180615 Interspeech-18 對話中的語音識别：Unsupervised Adaptation with Interpretable Disentangled Representations for Distant Conversational Speech Recognition
• 20180614 arXiv 跨資料集的person reid：Cross-dataset Person Re-Identification Using Similarity Preserved Generative Adversarial Networks
• 20180614 arXiv 将遷移學習應用于多個speaker的文字到語音：Transfer Learning from Speaker Verification to Multispeaker Text-To-Speech Synthesis
• 20180613 SIGIR-18 多任務學習用于推薦系統：Explainable Recommendation via Multi-Task Learning in Opinionated Text Data
• 20180613 CVPR-18 跨資料集的VQA：Cross-Dataset Adaptation for Visual Question Answering
• 20180612 ICASSP-18 遷移學習用于資源少的情感分類：Semi-supervised and Transfer learning approaches for low resource sentiment classification
• 20180612 KDD-18 多任務學習用于ICU病人資料挖掘：Learning Tasks for Multitask Learning: Heterogenous Patient Populations in the ICU
• 20180610 CEIG-17 将遷移學習用于插圖分類：Transfer Learning for Illustration Classification
• 20180610 BioNLP-18 将遷移學習用于病人實體分類：Embedding Transfer for Low-Resource Medical Named Entity Recognition: A Case Study on Patient Mobility
• 20180610 MICCAI-18 将遷移學習用于前列腺圖分類：Adversarial Domain Adaptation for Classification of Prostate Histopathology Whole-Slide Images
• 20180610 arXiv 遷移學習用于Coffee crop分類：A Comparative Study on Unsupervised Domain Adaptation Approaches for Coffee Crop Mapping
• 20180605 arXiv 遷移學習應用于胸X光片分割：Semantic-Aware Generative Adversarial Nets for Unsupervised Domain Adaptation in Chest X-ray Segmentation
• 20180604 arXiv 用CNN遷移學習進行硬化症檢測：One-shot domain adaptation in multiple sclerosis lesion segmentation using convolutional neural networks
• 20180530 MNRAS 用遷移學習檢測銀河星系兼并：Using transfer learning to detect galaxy mergers
• 20180529 arXiv 遷移學習用于表情識别：Meta Transfer Learning for Facial Emotion Recognition
• 20180524 KDD-18 用遷移學習方法進行人們的ID遷移：Learning and Transferring IDs Representation in E-commerce
• 20180519 arXiv 用遷移學習進行物體檢測，200幀/秒：Object detection at 200 Frames Per Second
• 20180519 arXiv 用遷移學習進行肢體語言識别：Optimization of Transfer Learning for Sign Language Recognition Targeting Mobile Platform
• 20180516 ACL-18 将對抗遷移學習用于危機狀态下的輿情分析：Domain Adaptation with Adversarial Training and Graph Embeddings
• 20180504 arXiv 用遷移學習進行心髒病檢測分類：ECG Heartbeat Classification: A Deep Transferable Representation
• 20180427 CVPR-18(workshop) 将深度遷移學習用于Person-reidentification： Adaptation and Re-Identification Network: An Unsupervised Deep Transfer Learning Approach to Person Re-Identification
• 20180426 arXiv 遷移學習用于醫學名字實體檢測；Label-aware Double Transfer Learning for Cross-Specialty Medical Named Entity Recognition
• 20180425 arXiv 将bagging和dropping結合起來進行遷移的一個深度網絡：A New Channel Boosted Convolution Neural Network using Transfer Learning
• 20180425 arXiv 遷移學習應用于自然語言任務：Dropping Networks for Transfer Learning
• 20180421 arXiv 采用聯合分布适配的深度遷移網絡用于工業生産中的錯誤診斷：Deep Transfer Network with Joint Distribution Adaptation: A New Intelligent Fault Diagnosis Framework for Industry Application
• 20180419 arXiv 跨領域的推薦系統：CoNet: Collaborative Cross Networks for Cross-Domain Recommendation
• 20180413 arXiv 跨模态檢索：Cross-Modal Retrieval with Implicit Concept Association
• 20180410 arXiv 用遷移學習進行犯罪現場的圖像比對：Cross-Domain Image Matching with Deep Feature Maps
• 20180408 ASRU-18 用遷移學習中的domain separation network進行speech recognition：Unsupervised Adaptation with Domain Separation Networks for Robust Speech Recognition
• 20180408 arXiv 小資料集上的遷移學習手寫體識别：Boosting Handwriting Text Recognition in Small Databases with Transfer Learning
• 20180404 arXiv 用遷移學習進行物體檢測：Transferring Common-Sense Knowledge for Object Detection
• 20180402 arXiv 将遷移學習用于癌症檢測：Improve the performance of transfer learning without fine-tuning using dissimilarity-based multi-view learning for breast cancer histology images
• 遷移學習用于行為識别 Transfer learning for activity recognition

https://arxiv.org/pdf/1411.1792.pdf

https://www.nature.com/articles/nature21056.epdf?referrer_access_token=_snzJ5POVSgpHutcNN4lEtRgN0jAjWel9jnR3ZoTv0NXpMHRAJy8Qn10ys2O4tuP9jVts1q2g1KBbk3Pd3AelZ36FalmvJLxw1ypYW0UxU7iShiMp86DmQ5Sh3wOBhXDm9idRXzicpVoBBhnUsXHzVUdYCPiVV0Slqf-Q25Ntb1SX_HAv3aFVSRgPbogozIHYQE3zSkyIghcAppAjrIkw1HtSwMvZ1PXrt6fVYXt-dvwXKEtdCN8qEHg0vbfl4_m&tracking_referrer=edition.cnn.com