ML paper notes

ML Papers

This repo contains notes and short summaries of some ML related papers I come across, organized by subjects and the summaries are in the form of PDFs.

Self-Supervised & Contrastive Learning

• Self-Supervised Relational Reasoning for Representation Learning (2020): [Paper] [Notes]
• Big Self-Supervised Models are Strong Semi-Supervised Learners (2020) [Paper] [Notes]
• Debiased Contrastive Learning (2020) [Paper] [Notes]
• Selfie: Self-supervised Pretraining for Image Embedding (2019): [Paper] [Notes]
• Self-Supervised Representation Learning by Rotation Feature Decoupling (2019): [Paper] [Notes]
• Revisiting Self-Supervised Visual Representation Learning (2019): [Paper] [Notes]
• AET vs. AED: Unsupervised Representation Learning by Auto-Encoding Transformations (2019): [Paper] [Notes]
• Boosting Self-Supervised Learning via Knowledge Transfer (2018): [Paper] [Notes]
• Self-Supervised Feature Learning by Learning to Spot Artifacts (2018): [Paper] [Notes]
• Unsupervised Representation Learning by Predicting Image Rotations (2018): [Paper] [Notes]
• Cross Pixel Optical-Flow Similarity for Self-Supervised Learning (2018): [Paper] [Notes]
• Multi-task Self-Supervised Visual Learning (2017): [Paper] [Notes]
• Split-Brain Autoencoders: Unsupervised Learning by Cross-Channel Prediction (2017): [Paper] [Notes]
• Colorization as a Proxy Task for Visual Understanding (2017): [Paper] [Notes]
• Unsupervised Learning of Visual Representations by Solving Jigsaw Puzzles (2017): [Paper] [Notes]
• Unsupervised Visual Representation Learning by Context Prediction (2016): [Paper] [Notes]
• Colorful image colorization (2016): [Paper] [Notes]
• Learning visual groups from co-occurrences in space and time (2015): [Paper] [Notes]
• Discriminative unsupervised feature learning with exemplar convolutional neural networks (2015): [Paper] [Notes]

Semi-Supervised Learning

• Negative sampling in semi-supervised learning (2020): [Paper] [Notes]
• Time-Consistent Self-Supervision for Semi-Supervised Learning (2020): [Paper] [Notes]
• Dual Student: Breaking the Limits of the Teacher in Semi-supervised Learning (2019): [Paper] [Notes]
• S4L: Self-Supervised Semi-Supervised Learning (2019): [Paper] [Notes]
• Semi-Supervised Learning by Augmented Distribution Alignment (2019): [Paper] [Notes]
• MixMatch: A Holistic Approach toSemi-Supervised Learning (2019): [Paper] [Notes]
• Unsupervised Data Augmentation (2019): [Paper] [Notes]
• Interpolation Consistency Training for Semi-Supervised Learning (2019): [Paper] [Notes]
• Deep Co-Training for Semi-Supervised Image Recognition (2018): [Paper] [Notes]
• Unifying semi-supervised and robust learning by mixup (2019): [Paper] [Notes]
• Realistic Evaluation of Deep Semi-Supervised Learning Algorithms (2018): [Paper] [Notes]
• Semi-Supervised Sequence Modeling with Cross-View Training (2018): [Paper] [Notes]
• Virtual Adversarial Training (2017): [Paper] [Notes]
• Mean teachers are better role models (2017): [Paper] [Notes]
• Temporal Ensembling for Semi-Supervised Learning (2017): [Paper] [Notes]
• Semi-Supervised Learning with Ladder Networks (2015): [Paper] [Notes]

Video Understanding

• Multiscale Vision Transformers (2021): [Paper] [Notes]
• ViViT A Video Vision Transformer (2021): [Paper] [Notes]
• Space-time Mixing Attention for Video Transformer (2021): [Paper] [Notes]
• Is Space-Time Attention All You Need for Video Understanding (2021): [Paper] [Notes]
• An Image is Worth 16x16 Words What is a Video Worth (2021): [Paper] [Notes]
• Temporal Query Networks for Fine-grained Video Understanding (2021): [Paper] [Notes]
• X3D Expanding Architectures for Efficient Video Recognition (2020): [Paper] [Notes]
• Temporal Pyramid Network for Action Recognition (2020): [Paper] [Notes]
• STM SpatioTemporal and Motion Encoding for Action Recognition (2019): [Paper] [Notes]
• Video Classification with Channel-Separated Convolutional Networks (2019): [Paper] [Notes]
• Video Modeling with Correlation Networks (2019): [Paper] [Notes]
• Videos as Space-Time Region Graphs (2018): [Paper] [Notes]
• SlowFast Networks for Video Recognition (2018): [Paper] [Notes]
• TSM Temporal Shift Module for Efficient Video Understanding (2018): [Paper] [Notes]
• Timeception for Complex Action Recognition (2018): [Paper] [Notes]
• Non-local Neural Networks (2017): [Paper] [Notes]
• Temporal Segment Networks for Action Recognition in Videos. (2017): [Paper] [Notes]
• Quo Vadis Action Recognition A New Model and the Kinetics Dataset (2017): [Paper] [Notes]
• A Closer Look at Spatiotemporal Convolutions for Action Recognition (2017): [Paper] [Notes]
• ActionVLAD Learning spatio-temporal aggregation for action classification (2017): [Paper] [Notes]
• Spatiotemporal Residual Networks for Video Action Recognition (2016): [Paper] [Notes]
• Deep Temporal Linear Encoding Networks (2016): [Paper] [Notes]
• Temporal Convolutional Networks for Action Segmentation and Detection (2016): [Paper] [Notes]
• Learning Spatiotemporal Features with 3D Convolutional Network (2014): [Paper] [Notes]

Domain Adaptation, Domain & Out-of-Distribution Generalization

• Rethinking Distributional Matching Based Domain Adaptation (2020): [Paper] [Notes]
• Transferability vs. Discriminability: Batch Spectral Penalization (2019): [Paper] [Notes]
• On Learning Invariant Representations for Domain Adaptation (2019): [Paper] [Notes]
• Universal Domain Adaptation (2019): [Paper] [Notes]
• Transferable Adversarial Training (2019): [Paper] [Notes]
• Multi-Adversarial Domain Adaptation (2018): [Paper] [Notes]
• Conditional Adversarial Domain Adaptation (2018): [Paper] [Notes]
• Learning Adversarially Fair and Transferable Representations (2018): [Paper] [Notes]
• What is the Effect of Importance Weighting in Deep Learning? (2018): [Paper] [Notes]

Explainability

• Towards Interpreting and Mitigating Shortcut Learning Behavior of NLU Models (2021): [Paper] [Notes]
• Transformer Interpretability Beyond Attention Visualization (2020): [Paper] [Notes]
• What shapes feature representations Exploring datasets architectures and training (2020): [Paper] [Notes]
• Attention-based Dropout Layer for Weakly Supervised Object Localization (2019): [Paper] [Notes]
• Attention is not Explanation (2019): [Paper] [Notes]
• SmoothGrad removing noise by adding noise (2017): [Paper] [Notes]
• Axiomatic Attribution for Deep Networks (2017): [Paper] [Notes]
• Attention Branch Network: Learning of Attention Mechanism for Visual Explanation (2019): [Paper] [Notes]
• Paying More Attention to Attention: Improving the Performance of CNNs via Attention Transfer (2016): [Paper] [Notes]

Natural Language Processing (NLP)

• Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing (2021): [Paper] [Notes]
• Unsupervised Data Augmentation with Naive Augmentation and without Unlabeled Data (2020): [Paper] [Notes]
• Supervised Contrastive Learning for Pre-trained Language Model Fine-tuning (2021): [Paper] [Notes]
• BERT and PALs: Projected Attention Layers for Efficient Adaptation in Multi-Task Learning (2020): [Paper] [Notes]
• FreeLB: Enhanced Adversarial Training for Natural Language Understanding (2020): [Paper] [Notes]
• MixText: Linguistically-Informed Interpolation for Semi-Supervised Text Classification (2020): [Paper] [Notes]

Generative Modeling

• Generative Pretraining from Pixels (2020): [Paper] [Notes]
• Consistency Regularization for Generative Adversarial Networks (2020): [Paper] [Notes]

Unsupervised Learning

• Invariant Information Clustering for Unsupervised Image Classification and Segmentation (2019): [Paper] [Notes]
• Deep Clustering for Unsupervised Learning of Visual Feature (2018): [Paper] [Notes]

Semantic Segmentation

• DeepLabv3+: Encoder-Decoder with Atrous Separable Convolution (2018): [Paper] [Notes]
• Large Kernel Matter, Improve Semantic Segmentation by Global Convolutional Network (2017): [Paper] [Notes]
• Understanding Convolution for Semantic Segmentation (2018): [Paper] [Notes]
• Rethinking Atrous Convolution for Semantic Image Segmentation (2017): [Paper] [Notes]
• RefineNet: Multi-path refinement networks for high-resolution semantic segmentation (2017): [Paper] [Notes]
• Pyramid Scene Parsing Network (2017): [Paper] [Notes]
• SegNet: A Deep ConvolutionalEncoder-Decoder Architecture for ImageSegmentation (2016): [Paper] [Notes]
• ENet: A Deep Neural Network Architecture for Real-Time Semantic Segmentation (2016): [Paper] [Notes]
• Attention to Scale: Scale-aware Semantic Image Segmentation (2016): [Paper] [Notes]
• Deeplab: semantic image segmentation with DCNN, atrous convs and CRFs (2016): [Paper] [Notes]
• U-Net: Convolutional Networks for Biomedical Image Segmentation (2015): [Paper] [Notes]
• Fully Convolutional Networks for Semantic Segmentation (2015): [Paper] [Notes]
• Hypercolumns for object segmentation and fine-grained localization (2015): [Paper] [Notes]

Weakly- and Semi-supervised Semantic segmentation

• Box-driven Class-wise Region Masking and Filling Rate Guided Loss (2019): [Paper] [Notes]
• FickleNet: Weakly and Semi-supervised Semantic Segmentation using Stochastic Inference (2019): [Paper] [Notes]
• Weakly-Supervised Semantic Segmentation Network with Deep Seeded Region Growing (2018): [Paper] [Notes]
• Learning Pixel-level Semantic Affinity with Image-level Supervision (2018): [Paper] [Notes]
• Object Region Mining with Adversarial Erasing (2018): [Paper] [Notes]
• Revisiting Dilated Convolution: A Simple Approach for Weakly- and Semi- Supervised Segmentation (2018): [Paper] [Notes]
• Tell Me Where to Look: Guided Attention Inference Network (2018): [Paper] [Notes]
• Semi Supervised Semantic Segmentation Using Generative Adversarial Network (2017): [Paper] [Notes]
• Decoupled Deep Neural Network for Semi-supervised Semantic Segmentation (2015): [Paper] [Notes]
• Weakly- and Semi-Supervised Learning of a DCNN for Semantic Image Segmentation (2015): [Paper] [Notes]

Information Retrieval

• VSE++: Improving Visual-Semantic Embeddings with Hard Negatives (2018): [Paper] [Notes]

Graph Neural Network

• Pixels to Graphs by Associative Embedding (2017): [Paper] [Notes]
• Associative Embedding: End-to-End Learning forJoint Detection and Grouping (2017): [Paper] [Notes]
• Interaction Networks for Learning about Objects , Relations and Physics (2016): [Paper] [Notes]
• DeepWalk: Online Learning of Social Representation (2014): [Paper] [Notes]
• The graph neural network model (2009): [Paper] [Notes]

Regularization

• Manifold Mixup: Better Representations by Interpolating Hidden States (2018): [Paper] [Notes]

Deep learning Methods & Models

• AutoAugment (2018): [Paper] [Notes]
• Stacked Hourgloass (2017): [Paper] [Notes]

Document analysis and segmentation

• dhSegment: A generic deep-learning approach for document segmentation (2018): [Paper] [Notes]
• Learning to extract semantic structure from documents using multimodal fully convolutional neural networks (2017): [Paper] [Notes]
• Page Segmentation for Historical Handwritten Document Images Using Conditional Random Fields (2016): [Paper] [Notes]
• ICDAR 2015 competition on text line detection in historical documents (2015): [Paper] [Notes]
• Handwritten text line segmentation using Fully Convolutional Network (2017): [Paper] [Notes]
• Deep Neural Networks for Large Vocabulary Handwritten Text Recognition (2015): [Paper] [Notes]
• Page Segmentation of Historical Document Images with Convolutional Autoencoders (2015): [Paper] [Notes]