MachineLearning-Papers_Survey

機械学習関連の論文 Survey 用レポジトリです。
論文まとめ記事は、Issues に記載しています。進捗は、Projects ページで管理しています。

■ 構成

◎ 基礎系（基礎モデル）

CNN

- [[ResNet] Deep Residual Learning for Image Recognition](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_NN_Note.md#ResNet%EF%BC%88%E6%AE%8B%E5%B7%AE%E3%83%8D%E3%83%83%E3%83%88%E3%83%AF%E3%83%BC%E3%82%AF%EF%BC%89) - [Spatial Transformer Networks](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/48)

GCN

- [Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_NN_Note.md#Convolutional_Neural_Networks_on_Graphs_with_Fast_Localized_Spectral_Filtering) - [Semi-Supervised Classification with Graph Convolutional Networks](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_NN_Note.md#Semi-Supervised_Classification_with_Graph_Convolutional_Networks) - [[R-GCN] Relational Graph Convolutional Network](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_NN_Note.md#R-GCN%EF%BC%88%E3%82%B0%E3%83%A9%E3%83%95%E3%83%95%E3%83%BC%E3%83%AA%E3%82%A8%E5%A4%89%E6%8F%9B%E3%82%92%E7%94%A8%E3%81%84%E3%81%AA%E3%81%84%E3%82%B0%E3%83%A9%E3%83%95%E7%95%B3%E3%81%BF%E8%BE%BC%E3%81%BF%EF%BC%89)

RNN

- [[RNN] Recursive Neural Network](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_NN_Note.md#%E3%83%AA%E3%82%AB%E3%83%AC%E3%83%B3%E3%83%88%E3%83%8B%E3%83%A5%E3%83%BC%E3%83%A9%E3%83%AB%E3%83%8D%E3%83%83%E3%83%88%E3%83%AF%E3%83%BC%E3%82%AF-rnn--recursive-neural-network%E9%9A%8E%E5%B1%A4%E5%9E%8B%E3%83%8B%E3%83%A5%E3%83%BC%E3%83%A9%E3%83%AB%E3%83%8D%E3%83%83%E3%83%88%E3%83%AF%E3%83%BC%E3%82%AF) - [[LSTM] long short-term memory](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_NN_Note.md#%E9%95%B7%E7%9F%AD%E6%9C%9F%E8%A8%98%E6%86%B6lstm-long-short-term-memory%E3%83%A2%E3%83%87%E3%83%AB) - [[GRU] gated recurrent unit](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_NN_Note.md#gru-gated-recurrent-unit)

Transformer

- [[Transformer] Attention Is All You Need](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/115) - [[Vision Transformer] An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/116) - [TransGAN: Two Transformers Can Make One Strong GAN](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/117)

VAE

- [[VAE] Auto-Encoding Variational Bayes](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#VAE) - [[VQ-VAE] Neural Discrete Representation Learning](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/23) - [β-VAE: LEARNING BASIC VISUAL CONCEPTS WITH A CONSTRAINED VARIATIONAL FRAMEWORK](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/27)

GANs

- [[GAN] Generative Adversarial Networks](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#GAN) - [[DCGAN] Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#DCGAN) - [[cGAN] Conditional Generative Adversarial Nets](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#ConditionalGAN%EF%BC%88cGAN%EF%BC%89) - [[WGAN] Wasserstein GAN](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#WGAN) - [[WGAN-gp] improved Training of Wasserstein GANs](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/25) - [SAGAN [Self-Attention Generative Adversarial Networks]](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#SAGAN) - [InfoGAN: Interpretable Representation Learning by Information Maximizing Generative Adversarial Nets](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/26) - [[RSGAN,RGAN,RaGAN] The relativistic discriminator: a key element missing from standard GAN](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/51) - [GAN-Tree: An Incrementally Learned Hierarchical Generative Framework for Multi-Modal Data Distributions](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/19) - [A U-Net Based Discriminator for Generative Adversarial Networks](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/110)

Flow-based generative model

- [NICE: NON-LINEAR INDEPENDENT COMPONENTS ESTIMATION](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/10) - [Real NVP [Density estimation using Real NVP]](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/11) - [Glow [Generative Flow with Invertible 1×1 Convolutions]](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/13) - [i-ResNets [Invertible residual networks]](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/14) - [Residual Flows for Invertible Generative Modeling](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/15)

Autoregressive Models

- [[PixelRNN, PixelCNN] Pixel Recurrent Neural Networks](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/24)

meta-learning, few-shot learning

- [MAML:Model Agnostic Meta-Learning for Fast Adaption](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/39)

Neural-ODE

- [[Neural-ODE] Neural Ordinary Differential Equations](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/21) - [Augmented Neural ODEs](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/37)

Neural Processes

- [Conditional Neural Processes](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/38) - [Neural Processes](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/44)

◎ アプリケーション系（CV）

Image Classification

- xxx

Semantic Segmentation

- [U-Net: Convolutional Networks for Biomedical Image Segmentation](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#UNet) - [[PSPNet] Pyramid Scene Parsing Network](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/69) - [Pyramid Attention Network for Semantic Segmentation](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/63) - [[DeepLab v3+] Encoder-Decoder with Atrous Separable Convolution for Semantic Image Segmentation](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/68) - [Concurrent Spatial and Channel ‘Squeeze & Excitation’ in Fully Convolutional Networks](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/62) - [Hypercolumns for Object Segmentation and Fine-grained Localization](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/64) - [Tversky loss function for image segmentation using 3D fully convolutional deep networks](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/67) - [Boundary loss for highly unbalanced segmentation](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/77) - Human Parsing - [[JPPNet] Look into Person: Joint Body Parsing & Pose Estimation Network and A New Benchmark](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/70) - [[CE2P] Devil in the Details: Towards Accurate Single and Multiple Human Parsing](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/72) - [Graphonomy: Universal Human Parsing via Graph Transfer Learning](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/8) - [Hierarchical Human Parsing with Typed Part-Relation Reasoning](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/79) - [[CorrPM] Correlating Edge, Pose with Parsing](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/74)

Object Detection

- [Fast R-CNN](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/75) - [Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/76) - [Focal Loss for Dense Object Detection](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/66)

Instance Segmentation

- [Mask R-CNN](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/65) - Human Parsing - [Parsing R-CNN for Instance-Level Human Analysis](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/73)

Pose Estimation

- [DensePose: Dense Human Pose Estimation in the Wild](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/50)

Image Registration / geometric matching

- [Convolutional neural network architecture for geometric matching](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/36)

image-to-image

- [[pix2pix] Image-to-Image Translation with Conditional Adversarial Networks](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#pix2pix) - [[pix2pix-HD] High-Resolution_Image_Synthesis_and_Semantic_Manipulation_with_Conditional_GANs](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/18) - [[CycleGAN] Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#CycleGAN) - [[StarGAN] StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#StarGAN) - [[SPADE] Semantic Image Synthesis with Spatially-Adaptive Normalization](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/7) - [[Neural Collage] Spatially Controllable Image Synthesis with Internal Representation Collaging](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/22) - [Recapture as You Want](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/78) - [Encoding in Style: a StyleGAN Encoder for Image-to-Image Translation](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/104) - [[Impersonator++] Liquid Warping GAN with Attention: A Unified Framework for Human Image Synthesis](https://github.com/Yagami360/MachineLearning-Papers_Survey_Private/issues/5) - [Focal Frequency Loss for Generative Models](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/114) - few-shot learning - [SinGAN: Learning a Generative Model from a Single Natural Image](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/40) - [[DeepSIM] Deep Single Image Manipulation](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/107)

noize-to-image

- [[PGGAN] Progressive Growing of GANs for Improved Quality, Stability, and Variation](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#ProgressiveGAN%EF%BC%88PGGAN%EF%BC%89) - [[StyleGAN] A Style-Based Generator Architecture for Generative Adversarial Networks](https://github.com/Yagami360/My_NoteBook/blob/master/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6/%E6%83%85%E5%A0%B1%E5%B7%A5%E5%AD%A6_%E6%A9%9F%E6%A2%B0%E5%AD%A6%E7%BF%92_%E7%94%9F%E6%88%90%E3%83%A2%E3%83%87%E3%83%AB.md#StyleGAN) - [[StyleGAN2] Analyzing and Improving the Image Quality of StyleGAN](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/52) - few-shot learning - [Training Generative Adversarial Networks with Limited Data / StyleGAN2 with adaptive discriminator augmentation (ADA)](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/108) - [Data-Efficient GANs with DiffAugment](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/109) - [[Lightweight GAN] Towards Faster and Stabilized GAN Training for High-fidelity Few-shot Image Synthesis](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/113)

Inpainting

- [[Deepfillv2] Free-Form Image Inpainting with Gated Convolution](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/29) - [Pluralistic Image Completion](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/6) - [Boundless: Generative Adversarial Networks for Image Extension](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/28)

Person Image Generation

- [Pose Guided Person Image Generation](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/49) - [Disentangled Person Image Generation](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/41) - [[Soft-Gated Warping-GAN] Soft-Gated Warping-GAN for Pose-Guided Person Image Synthesis](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/35)

recommendation

- [ViBE: Dressing for Diverse Body Shapes](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/60)

text-to-image

- [[StackGAN] Text to Photo-realistic Image Synthesis with Stacked Generative Adversarial Network](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/9)

Optical Flow

- [FlowNet: Learning Optical Flow with Convolutional Networks](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/43) - [View Synthesis by Appearance Flow](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/47)

3D Reconstruction

- param-to-3D / parametric 3D models - [SMPL: A skinned multi-person linear model ](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/86) - [[CAPE] Learning to Dress 3D People in Generative Clothing](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/93) - [TailorNet: Predicting Clothing in 3D as a Function of Human Pose, Shape and Garment Style](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/91) - [SIZER: A Dataset and Model for Parsing 3D Clothing and Learning Size Sensitive 3D Clothing](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/84) - image-to-3D / image-based 3D Reconstruction - none-parametric 3D models - [Mesh R-CNN](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/83) - [Occupancy Networks: Learning 3D Reconstruction in Function Space](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/90) - [PIFu: Pixel-Aligned Implicit Function for High-Resolution Clothed Human Digitization](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/99) - [PIFuHD: Multi-Level Pixel-Aligned Implicit Function for High-Resolution 3D Human Digitization](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/94) - [NormalGAN: Learning Detailed 3D Human from a Single RGB-D Image](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/95) - using templete mesh - [3D Virtual Garment Modeling from RGB Images](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/81) - parametric 3D models - [[HMR] End-to-end Recovery of Human Shape and Pose](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/98) - [Multi-Garment Net: Learning to Dress 3D People from Images](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/87) - [Deep Fashion3D: A Dataset and Benchmark for 3D Garment Reconstruction from Single Images](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/89) - [BCNet: Learning Body and Cloth Shape from A Single Image](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/92) - [ExPose: Monocular Expressive Body Regression through Body-Driven Attention](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/97) - [I2L-MeshNet: Image-to-Lixel Prediction Network for Accurate 3D Human Pose and Mesh Estimation from a Single RGB Image](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/100) - video-to-3D - [TexMesh: Reconstructing Detailed Human Texture and Geometry from RGB-D Video](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/96) - texture mapping - [360-Degree Textures of People in Clothing from a Single Image](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/103) - [Learning to Transfer Texture from Clothing Images to 3D Humans](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/85) - camera localization - [PoseNet: A Convolutional Network for Real-Time 6-DOF Camera Relocalization](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/101) - [Geometry-Aware Learning of Maps for Camera Localization](https://github.com/Yagami360/MachineLearning-Papers_Survey/issues/102)

Others

- xxx

◎ 理論系

xxx

- xxx

■ 機械学習系の論文調査や論文の読み方

論文調査
- その論文の発展版を探すには、Google Scholer でその論文の被引用論文を探すのが手っ取り早い。
- 基本的には、その論文の被引用数が多いものが注目の論文。但し最新の論文は当然被引用数は少なくなる。
- 著者の所属組織も優れた論文であるか判断材料になり得る。例えば、GAFA や NVIDIA, Adobe などの論文は、優れた論文である可能性が高い
- 基本的には、最新の論文ほど優れた実験結果を出しているので、Google Scholer で出来るだけ公開日時の最新のものを探す。
- 論文の公式実装がないものは、自前実装の再現コストが高いので、公式実装があるものを優先的に探す。
- 論文の公式実装の有無は、papers with code で確認できる。
- github 上に論文公式実装がある場合は、スター数も優れた論文であるかの判断材料になり得る。
- 探している分野の論文が１つ見つかれば、その論文を papers with code で検索する。その論文の "Tasks" が割り振られている場合は、その Tasks の名前からその分野の最新論文を調べることが出来る。
論文の読み方
- 基本的に、「Abstract」→「Introduction」→「Conclusion」→「何をしたかの詳細」→「Experiments」→「Related Work」項目の順に読むのが効率的。Related Work は最悪飛ばしても良い。
- Abstract や Introduction 内の "In this paper ...", "In this work ..." という文面や、Introduction の最後にある "the contributions of this paper ... as follows." という文面には注目。
- その論文の未解決課題や問題点は、論文中の Future work や Conculution に述べられていることが多い。或いは、その論文を引用している論文の Related Work に述べられていることが多い。

■ 論文要約フォーマット（要約バージョン）

layout: post

title:  "論文タイトル"

date:   YYYY-MM-DD

categories: CV NLP Others

## 1. どんなもの？

## 2. 先行研究と比べてどこがすごいの？

## 3. 技術や手法の"キモ"はどこにある？

![Figure 1]({{ site.baseurl }}/assets/img/(cv, nlp, others)/(title)/figure1.png)

## 4. どうやって有効だと検証した？

## 5. 議論はあるか？

## 6. 次に読むべき論文はあるか？

### 論文情報・リンク

* [著者，"タイトル，" ジャーナル名，voluem，no.，ページ，年](論文リンク)

Yagami360 / machine-learning-papers-survey

readme

MachineLearning-Papers_Survey

■ 構成

◎ 基礎系（基礎モデル）

◎ アプリケーション系（CV）

◎ 理論系

■ 機械学習系の論文調査や論文の読み方

■ 論文要約フォーマット（要約バージョン）

■ 参考サイト

◎ 論文サイト

◎ 便利サイト

◎ その他参考サイト