点击上方“视学算法”，选择加"星标"或“置顶”
重磅干货，第一时间送达
转自：新智元
好像很多人都觉得读论文做笔记是一件非常正确、重要、且必要的事情。你可能没少看有关《为什么习惯记笔记的人更容易成功》《学霸的笔记都是这么做的，难怪他们的考试分数那么高！》等文章。
论文笔记的作用不仅仅是起到一种心理安慰，更重要的是划出重点、标记出知识盲区，便于后续温习。德国心理学家艾宾浩斯研究发现，人类在学习后就会开始遗忘，遗忘的程度是不均匀的，刚开始记忆下降比例很高，后面会越来越少。根据实验结果发现刚记住的时候是100%，过了20分钟记忆程度只有58.2%，2天和6天后分别只能记得27.8%和25.4%。如果只是单纯的看论文，很难将里面的知识点据为己有，这个时候就需要笔记来帮忙来延缓知识的流失速度了。我们经常看到网上动不动就最全XX论文全集，上来就是几十篇论文。人类都有囤积物资的欲求，囤积论文也差不多。我们往往觉得，我手中的论文数量越多，带给我的安全感就越大、我能收获的知识就越多。虽然理智告诉我们：这样想法是错误的！然而欲望却拖着我们像个过冬的松鼠一样，不断的收集松果。但如果硬是强迫自己看一篇论文就必须要做多少多少笔记，也不太现实。而且如果一开始没有培养期做笔记习惯的话，很可能一开始并没有get到做笔记的精髓，导致事无巨细全都记笔记。这个时候，看看别人做的笔记，尤其是看看学霸做的笔记，也是一个非常不错、极具实操性的方法。今天新智元为大家带来一位博士生的论文笔记。这位学霸真的是有很认真的看了不少论文，为了便于检索，他在看过的ML相关的论文进行了注释和简短摘要，并且将这些摘要按主题分类。他将论文和笔记都放在了GitHub上，非常方便进行对照。以下就是论文和笔记的列表，大家可以根据需要下载阅读。Self-Supervised Learning
论文标题：Selfie: Self-supervised Pretraining for Image Embedding (2019)论文链接：https://arxiv.org/abs/1906.02940笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/76_selfie_pretraining_for_img_embeddings.pdf论文标题：Self-Supervised Representation Learning by Rotation Feature Decoupling (2019)论文链接：https://github.com/philiptheother/FeatureDecoupling笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/73_SSL_by_rotation_decoupling.pdf论文标题：Revisiting Self-Supervised Visual Representation Learning (2019)论文链接：https://arxiv.org/abs/1901.09005笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/72_revisiting_SSL.pdf论文标题：AET vs. AED: Unsupervised Representation Learning by Auto-Encoding Transformations rather than Data (2019)论文链接：https://arxiv.org/abs/1901.04596笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/74_AFT_vs_AED.pdf论文标题：Boosting Self-Supervised Learning via Knowledge Transfer (2018)论文链接：https://arxiv.org/abs/1805.00385笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/67_boosting_self_super_via_trsf_learning.pdf论文标题：Self-Supervised Feature Learning by Learning to Spot Artifacts (2018)论文链接：https://arxiv.org/abs/1806.05024笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/69_SSL_by_learn_to_spot_artifacts.pdf论文标题：Unsupervised Representation Learning by Predicting Image Rotations (2018)论文链接：https://arxiv.org/abs/1803.07728笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/68_unsup_img_rep_learn_by_rot_predic.pdf论文标题：Cross Pixel Optical-Flow Similarity for Self-Supervised Learning (2018)论文链接：https://arxiv.org/abs/1807.05636笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/75_cross_pixel_optical_flow.pdf论文标题：Multi-task Self-Supervised Visual Learning (2017)论文链接：https://arxiv.org/abs/1708.07860笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/64_multi_task_self_supervised.pdf论文标题：Split-Brain Autoencoders: Unsupervised Learning by Cross-Channel Prediction (2017)论文链接：https://arxiv.org/abs/1611.09842笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/65_split_brain_autoencoders.pdf论文标题：Colorization as a Proxy Task for Visual Understanding (2017)论文链接：https://arxiv.org/abs/1703.04044笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/66_colorization_as_a_proxy_for_viz_under.pdf论文标题：Unsupervised Learning of Visual Representations by Solving Jigsaw Puzzles (2017)论文链接：https://arxiv.org/abs/1603.09246笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/63_solving_jigsaw_puzzles.pdf论文标题：Unsupervised Visual Representation Learning by Context Prediction (2016)论文链接：https://arxiv.org/abs/1505.05192笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/62_unsupervised_learning_with_context_prediction.pdf论文标题：Colorful image colorization (2016)论文链接：https://richzhang.github.io/colorization/笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/59_colorful_colorization.pdf论文标题：Learning visual groups from co-occurrences in space and time (2015)论文链接：https://arxiv.org/abs/1511.06811笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/61_visual_groups_from_co_occurrences.pdf论文标题：Discriminative unsupervised feature learning with exemplar convolutional neural networks (2015)论文链接：https://arxiv.org/abs/1406.6909笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/60_exemplar_CNNs.pdfSemi-Supervised Learning论文标题：Dual Student: Breaking the Limits of the Teacher in Semi-supervised Learning (2019)论文链接：https://arxiv.org/abs/1909.01804笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/79_dual_student.pdf论文标题：S4L: Self-Supervised Semi-Supervised Learning (2019)论文链接：https://arxiv.org/abs/1905.03670笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/83_S4L.pdf论文标题：Semi-Supervised Learning by Augmented Distribution Alignment (2019)论文链接：https://arxiv.org/abs/1905.08171笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/80_SSL_aug_dist_align.pdf论文标题：MixMatch: A Holistic Approach toSemi-Supervised Learning (2019)论文链接：https://arxiv.org/abs/1905.02249笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/45_mixmatch.pdf论文标题：Unsupervised Data Augmentation (2019)论文链接：https://arxiv.org/abs/1904.12848笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/39_unsupervised_data_aug.pdf论文标题：Interpolation Consistency Training forSemi-Supervised Learning (2019)论文链接：https://arxiv.org/abs/1903.03825笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/44_interpolation_consistency_tranining.pdf论文标题：Deep Co-Training for Semi-Supervised Image Recognition (2018)论文链接：https://arxiv.org/abs/1803.05984笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/46_deep_co_training_img_rec.pdf论文标题：Unifying semi-supervised and robust learning by mixup (2019)论文链接：https://openreview.net/forum?id=r1gp1jRN_4笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/42_mixmixup.pdf论文标题：Realistic Evaluation of Deep Semi-Supervised Learning Algorithms (2018)论文链接：https://arxiv.org/abs/1804.09170笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/37_realistic_eval_of_deep_ss.pdf论文标题：Semi-Supervised Sequence Modeling with Cross-View Training (2018)论文链接：https://arxiv.org/abs/1809.08370笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/38_cross_view_semi_supervised.pdf论文标题：Virtual Adversarial Training:A Regularization Method for Supervised andSemi-Supervised Learning (2017)论文链接：https://arxiv.org/abs/1704.03976笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/40_virtual_adversarial_training.pdf论文标题：Mean teachers are better role models (2017)论文链接：https://arxiv.org/abs/1703.01780笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/56_mean_teachers.pdf论文标题：Temporal Ensembling for Semi-Supervised Learning (2017)论文链接：https://arxiv.org/abs/1610.02242笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/55_temporal-ensambling.pdf论文标题：Semi-Supervised Learning with Ladder Networks (2015)论文链接：https://arxiv.org/abs/1507.02672笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/33_ladder_nets.pdfUnsupervised Learning论文标题：Invariant Information Clustering for Unsupervised Image Classification and Segmentation (2019)论文链接：https://arxiv.org/abs/1807.06653笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/78_IIC.pdf论文标题：Deep Clustering for Unsupervised Learning of Visual Feature (2018)论文链接：https://arxiv.org/abs/1807.05520笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/70_deep_clustering_for_un_visual_features.pdf
Semantic Segmentation论文标题：DeepLabv3+: Encoder-Decoder with Atrous Separable Convolution (2018)论文链接：https://arxiv.org/abs/1802.02611笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/26_deeplabv3+.pdf论文标题：Large Kernel Matter, Improve Semantic Segmentation by Global Convolutional Network (2017)论文链接：https://arxiv.org/abs/1703.02719笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/28_large_kernel_maters.pdf论文标题：Understanding Convolution for Semantic Segmentation (2018)论文链接：https://arxiv.org/abs/1702.08502笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/29_understanding_conv_for_sem_seg.pdf论文标题：Rethinking Atrous Convolution for Semantic Image Segmentation (2017)论文链接：https://arxiv.org/abs/1706.05587笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/25_deeplab_v3.pdf论文标题：RefineNet: Multi-path refinement networks for high-resolution semantic segmentation (2017)论文链接：https://arxiv.org/abs/1611.06612笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/31_refinenet.pdf论文标题：Pyramid Scene Parsing Network (2017)论文链接：http://jiaya.me/papers/PSPNet_cvpr17.pdf笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/22_pspnet.pdf论文标题：SegNet: A Deep ConvolutionalEncoder-Decoder Architecture for ImageSegmentation (2016)论文链接：https://arxiv.org/pdf/1511.00561笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/21_segnet.pdf论文标题：ENet: A Deep Neural Network Architecture for Real-Time Semantic Segmentation (2016)论文链接：https://arxiv.org/abs/1606.02147笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/27_enet.pdf论文标题：Attention to Scale: Scale-aware Semantic Image Segmentation (2016)论文链接：https://arxiv.org/abs/1511.03339笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/30_atttention_to_scale.pdf论文标题：Deeplab: semantic image segmentation with DCNN, atrous convs and CRFs (2016)论文链接：https://arxiv.org/abs/1606.00915笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/23_deeplab_v2.pdf论文标题：U-Net: Convolutional Networks for Biomedical Image Segmentation (2015)论文链接：https://arxiv.org/abs/1505.04597笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/20_Unet.pdf论文标题：Fully Convolutional Networks for Semantic Segmentation (2015)论文链接：https://people.eecs.berkeley.edu/~jonlong/long_shelhamer_fcn.pdf笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/19_FCN.pdf论文标题：Hypercolumns for object segmentation and fine-grained localization (2015)论文链接：http://home.bharathh.info/pubs/pdfs/BharathCVPR2015.pdf笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/24_hypercolumns.pdfWeakly
论文标题：Box-driven Class-wise Region Masking and Filling Rate Guided Loss for Weakly Supervised Semantic Segmentation (2019)
论文链接：http://arxiv.org/abs/1904.11693笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/54_boxe_driven_weakly_segmentation.pdf论文标题：FickleNet: Weakly and Semi-supervised Semantic Image Segmentation using Stochastic Inference (2019)论文链接：https://arxiv.org/abs/1902.10421笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/49_ficklenet.pdf论文标题：Weakly-Supervised Semantic Segmentation Network with Deep Seeded Region Growing (2018)论文链接：http://openaccess.thecvf.com/content_cvpr_2018/papers/Huang_Weakly-Supervised_Semantic_Segmentation_CVPR_2018_paper.pdf笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/53_deep_seeded_region_growing.pdf论文标题：Learning Pixel-level Semantic Affinity with Image-level Supervision for Weakly Supervised Semantic Segmentation (2018)论文链接：https://arxiv.org/abs/1803.10464笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/81_affinity_for_ws_segmentation.pdf论文标题：Object Region Mining with Adversarial Erasing: A Simple Classification to Semantic Segmentation Approach (2018)论文链接：https://arxiv.org/abs/1703.08448笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/51_object_region_manning_for_sem_seg.pdf论文标题：Revisiting Dilated Convolution: A Simple Approach for Weakly论文标题：and Semi论文标题：Supervised Semantic Segmentation (2018)论文链接：https://arxiv.org/abs/1805.04574笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/52_dilates_convolution_semi_super_segmentation.pdf论文标题：Tell Me Where to Look: Guided Attention Inference Network (2018)论文链接：https://arxiv.org/abs/1802.10171笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/50_tell_me_where_to_look.pdf论文标题：Semi Supervised Semantic Segmentation Using Generative Adversarial Network (2017)论文链接：https://arxiv.org/abs/1703.09695笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/82_ss_segmentation_gans.pdf论文标题：Decoupled Deep Neural Network for Semi-supervised Semantic Segmentation (2015)论文链接：https://arxiv.org/abs/1506.04924笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/47_decoupled_nn_for_segmentation.pdf论文标题：Weakly论文标题：and Semi-Supervised Learning of a DCNN for Semantic Image Segmentation (2015)论文链接：https://arxiv.org/abs/1502.02734笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/48_weakly_and_ss_for_segmentation.pdfInformation Retrieval论文标题：VSE++: Improving Visual-Semantic Embeddings with Hard Negatives (2018)论文链接：https://arxiv.org/abs/1707.05612笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/77_vse++.pdf
Visual Explanation & Attention论文标题：Attention Branch Network: Learning of Attention Mechanism for Visual Explanation (2019)论文链接：https://arxiv.org/abs/1812.10025笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/57_attention_branch_netwrok.pdf论文标题：Attention-based Dropout Layer for Weakly Supervised Object Localization (2019)论文链接：http://openaccess.thecvf.com/content_CVPR_2019/papers/Choe_Attention-Based_Dropout_Layer_for_Weakly_Supervised_Object_Localization_CVPR_2019_paper.pdf笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/58_attention_based_dropout.pdf论文标题：Paying More Attention to Attention: Improving the Performance of Convolutional Neural Networks via Attention Transfer (2016)论文链接：https://arxiv.org/abs/1612.03928笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/71_attention_transfer.pdfGraph neural network & Graph embeddings论文标题：Pixels to Graphs by Associative Embedding (2017)论文链接：https://arxiv.org/abs/1706.07365笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/36_pixels_to_graphs.pdf论文标题：Associative Embedding: End-to-End Learning forJoint Detection and Grouping (2017)论文链接：https://arxiv.org/abs/1611.05424笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/35_associative_emb.pdf论文标题：Interaction Networks for Learning about Objects , Relations and Physics (2016)论文链接：https://arxiv.org/abs/1612.00222笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/18_interaction_nets.pdf论文标题：DeepWalk: Online Learning of Social Representation (2014)论文链接：http://www.perozzi.net/publications/14_kdd_deepwalk.pdf笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/deep_walk.pdf论文标题：The graph neural network model (2009)论文链接：https://persagen.com/files/misc/scarselli2009graph.pdf笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/graph_neural_nets.pdfRegularization
论文标题：Manifold Mixup: Better Representations by Interpolating Hidden States (2018)论文链接：https://arxiv.org/abs/1806.05236笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/43_manifold_mixup.pdfDeep learning Methods & Models
论文标题：AutoAugment (2018)
论文链接：https://arxiv.org/abs/1805.09501笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/41_autoaugment.pdf论文标题：Stacked Hourgloass (2017)论文链接：http://ismir2018.ircam.fr/doc/pdfs/138_Paper.pdf笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/34_stacked_hourglass.pdfDocument analysis and segmentation论文标题：dhSegment: A generic deep-learning approach for document segmentation (2018)论文链接：https://arxiv.org/abs/1804.10371笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/dhSegement.pdf论文标题：Learning to extract semantic structure from documents using multimodal fully convolutional neural networks (2017)论文链接：https://arxiv.org/abs/1706.02337笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/learning_to_extract.pdf论文标题：Page Segmentation for Historical Handwritten Document Images Using Conditional Random Fields (2016)论文链接：https://www.researchgate.net/publication/312486501_Page_Segmentation_for_Historical_Handwritten_Document_Images_Using_Conditional_Random_Fields笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/seg_with_CRFs.pdf论文标题：ICDAR 2015 competition on text line detection in historical documents (2015)论文链接：http://ieeexplore.ieee.org/abstract/document/7333945/笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/ICDAR2015.pdf论文标题：Handwritten text line segmentation using Fully Convolutional Network (2017)论文链接：https://ieeexplore.ieee.org/document/8270267/笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/handwritten_text_seg_FCN.pdf论文标题：Deep Neural Networks for Large Vocabulary Handwritten Text Recognition (2015)论文链接：https://tel.archives-ouvertes.fr/tel-01249405/document笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/andwriten_text_recognition.pdf论文标题：Page Segmentation of Historical Document Images with Convolutional Autoencoders (2015)论文链接：https://ieeexplore.ieee.org/abstract/document/7333914/笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/segmentation_with_CAE.pdf论文标题：A typed and handwritten text block segmentation system for heterogeneous and complex documents (2012)论文链接：https://www.researchgate.net/publication/275518176_A_Typed_and_Handwritten_Text_Block_Segmentation_System_for_Heterogeneous_and_Complex_Documents笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/a_typed_block_seg.pdf论文标题：Document layout analysis, Classical approaches (1992:2001)论文链接：https://pdfs.semanticscholar.org/5392/90b571b918da959fabaae7f605bb07850518.pdf笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/old_classical_approaches.pdf论文标题：Page Segmentation for Historical Document Images Based on Superpixel Classification with Unsupervised Feature Learning (2016)论文链接：https://ieeexplore.ieee.org/document/7490134笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/seg_with_superpixels.pdf论文标题：Paragraph text segmentation into lines with Recurrent Neural Networks (2015)论文链接：http://ieeexplore.ieee.org/abstract/document/7333803/笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/textlines_srg_with_RNNs.pdf论文标题：A comprehensive survey of mostly textual document segmentation algorithms since 2008 (2017 )论文链接：https://hal.archives-ouvertes.fr/hal-01388088/document笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/survey_doc_segmentation.pdf论文标题：Convolutional Neural Networks for Page Segmentation of Historical Document Images (2017)论文链接：https://arxiv.org/abs/1704.01474笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/CNNs_chen.pdf论文标题：ICDAR2009 Page Segmentation Competition (2009)论文链接：https://ieeexplore.ieee.org/document/5277763笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/ICDAR2009.pdf论文标题：Amethod for combining complementary techniques for document image segmentation (2009)论文链接：https://www.researchgate.net/publication/220600948_A_method_for_combining_complementary_techniques_for_document_image_segmentation笔记链接：https://github.com/yassouali/ML_paper_notes/blob/master/notes/a_method_for_combining_complementary_techniques.pdf
完整列表：https://github.com/yassouali/ML_paper_notes
本文暂时没有评论，来添加一个吧(●'◡'●)