Update to 2023-06-30.

README.md

@@ -145,6 +145,8 @@ There have been several reviews / surveys / benchmarks for this topic.
 ### **Benchmarks - 基线**:
 - [A Comparative Survey: Benchmarking for Pool-based Active Learning [2021, IJCAI]](https://www.ijcai.org/proceedings/2021/0634.pdf)
 - [A Framework and Benchmark for Deep Batch Active Learning for Regression [2022]](https://arxiv.org/pdf/2203.09410.pdf)
+- [Re-Benchmarking Pool-Based Active Learning for Binary Classification [2023]](https://arxiv.org/pdf/2306.08954.pdf)
+- [LabelBench: A Comprehensive Framework for Benchmarking Label-Efficient Learning [2023]](https://arxiv.org/pdf/2306.09910.pdf)
 
 ### **Tutorials - 教程**
 

conference/ICML.md

@@ -1,3 +1,5 @@
+- Discover-Then-Rank Unlabeled Support Vectors in the Dual Space for Multi-Class Active Learning [2023, ICML]
+- Towards Controlled Data Augmentations for Active Learning [2023, ICML]
 - N-Penetrate: Active Learning of Neural Collision Handler for Complex 3D Mesh Deformations [2022, ICML]
 - Active Multi-Task Representation Learning [2022, ICML]
 - ActiveHedge: Hedge meets Active Learning [2022, ICML]

conference/NeuraIPS.md

@@ -1,5 +1,6 @@
 
 - Batch Active Learning from the Perspective of Sparse Approximation [2022, NeurIPS]
+- LOG: Active Model Adaptation for Label-Efficient OOD Generalization [2022, NeuraIPS]
 - Few-Shot Continual Active Learning by a Robot [2022, NeurIPS]
 - [Corruption Robust Active Learning [2021, NeurIPS]](https://arxiv.org/pdf/2106.11220.pdf): 
 - [BatchBALD: Efficient and Diverse Batch Acquisition for Deep Bayesian Active Learning [2019, NeurIPS]](http://papers.nips.cc/paper/8925-batchbald-efficient-and-diverse-batch-acquisition-for-deep-bayesian-active-learning): 

contents/AL_combinations.md

@@ -172,6 +172,7 @@ Video Action Recognition:
 
 3D Semantic Segmentation:
 - LiDAL: Inter-frame Uncertainty Based Active Learning for 3D LiDAR Semantic Segmentation [2022, ECCV]
+- Efficient 3D Scene Semantic Segmentation via Active Learning on Rendered 2D Images [2023, TIP]
 
 Visual Tracking:
 - Active Learning for Deep Visual Tracking [2021]
@@ -200,9 +201,11 @@ Micro-Expression Recognition:
 - Tackling Micro-Expression Data Shortage via Dataset Alignment and Active Learning [2022, IEEE Trans Multimedia]
 
 Collision Prediction & Handling:
-
 - N-Penetrate: Active Learning of Neural Collision Handler for Complex 3D Mesh Deformations [2022, ICML]
 
+Super-resolution:
+- USIM-DAL: Uncertainty-aware Statistical Image Modeling-based Dense Active Learning for Super-resolution [2023, UAI]
+
 ## Natural Language Processing (NLP)
 
 NLP is also quite a wide conception.
@@ -299,6 +302,9 @@ Abstractive Text Summarization:
 Natural Language Explanations：
 - Beyond Labels: Empowering Human with Natural Language Explanations through a Novel Active-Learning Architecture [2023]
 
+n-ary relation extraction:
+- Active Learning for Cross-Sentence n-ary Relation Extraction [2023, Information Sciences]
+
 ## Domain adaptation/Transfer learning
 
 Normally when we use AL in domain adaptation, we can obtain several true labels of the unlabeled instances on source/target domain.
@@ -345,6 +351,8 @@ Transfer learning:
 - Active learning with cross-class similarity transfer [2017, AAAI]
 - Rapid Performance Gain through Active Model Reuse [IJCAI, 2019]
 
+Domain Generalization:
+- LOG: Active Model Adaptation for Label-Efficient OOD Generalization [2022, NeuraIPS]
 
 ## One/Few/Zero-shot learning or Meta-Learning
 
@@ -371,6 +379,7 @@ For one/few-shot learning:
 - MEAL: Stable and Active Learning for Few-Shot Prompting [2022]
 - Few-shot initializing of Active Learner via Meta-Learning [2022, EMNLP]
 - Active Learning for Efficient Few-Shot Classification [2023, ICASSP]
+- Improved prototypical network for active few-shot learning [2023, PRL]
 
 There are also works about zero-shot learning:
 - Graph active learning for GCN-based zero-shot classification [2021, Neurocomputing]
@@ -533,6 +542,7 @@ For example, the ImageNet was crawled from image databases without considering s
 
 - [Root Cause Analysis for Self-organizing Cellular Network: an Active Learning Approach](https://link.springer.com/article/10.1007/s11036-020-01589-1)
 - [Active Invariant Causal Prediction: Experiment Selection through Stability](https://arxiv.org/pdf/2006.05690.pdf)
+- ACE: Active Learning for Causal Inference with Expensive Experiments [2023]
 
 ## Choice Model
 
@@ -621,6 +631,7 @@ Normally there is only model selection without training.
 - [Deep Multi-Fidelity Active Learning of High-Dimensional Outputs [2020]](https://arxiv.org/pdf/2012.00901.pdf)
 - [Batch Multi-Fidelity Active Learning with Budget Constraints [2022, NeruaIPS]](https://arxiv.org/pdf/2210.12704.pdf)
 - Disentangled Multi-Fidelity Deep Bayesian Active Learning [2023]
+- Multi-Fidelity Active Learning with GFlowNets [2023]
 
 ## Online Learning System
 

contents/AL_technique_models.md

@@ -5,8 +5,8 @@ In this chapter, we care about how to apply AL on specific models.
 # Models
 
 ## SVM/LR
-Most common models, we won't waste time here.
-Most of classic strategies are based on these models.
+
+- Discover-Then-Rank Unlabeled Support Vectors in the Dual Space for Multi-Class Active Learning [2023, ICML]
 
 ## Bayesian/Probabilistic/Gaussian Progress
 - Employing EM and Pool-Based Active Learning for Text Classification [[1998. ICML]](http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.50.10&rep=rep1&type=pdf): 

contents/MVAL.md

@@ -6,4 +6,5 @@ AL select instances by considering the relations among views.
 Works:
 - Active + Semi-Supervised Learning = Robust Multi-View Learning [2002, ICML]
 - Active learning with multiple views [2006, JAIR]
-- Multi-view active learning in the non-realizable case [2010, NeurIPS]
+- Multi-view active learning in the non-realizable case [2010, NeurIPS]
+- Towards Balanced Active Learning for Multimodal Classification [2023]

contents/deep_AL.md

@@ -214,6 +214,7 @@ Data Augmentation:
   LADA takes the influence of the augmented data into account to construct acquisition function
 - Collaborative Intelligence Orchestration: Inconsistency-Based Fusion of Semi-Supervised Learning and Active Learning [KDD, 2022]
 - When Active Learning Meets Implicit Semantic Data Augmentation [2022, ECCV]
+- Towards Controlled Data Augmentations for Active Learning [2023, ICML]
 
 Labeled-unlabeled data indistinguishable:
 - [Deep Active Learning: Unified and Principled Method for Query and Training [2020, ICAIS]](https://arxiv.org/abs/1911.09162): WAAL.
@@ -244,4 +245,5 @@ Pretrain:
 
 Both self and semi-supervised imparted:
 - Self-supervised Semi-supervised Learning for Data Labeling and Quality Evaluation [2021, Arxiv]: 
-  Use random selection with BYOL and Label Propagation.
+  Use random selection with BYOL and Label Propagation.
+- Representation-Based Time Series Label Propagation for Active Learning [2023, CSCWD]

on_build/AL_theory.md

@@ -88,6 +88,10 @@ The covariate X might be the result of a complex process that the learner can in
 
 - Nonparametric Indirect Active Learning [2022, ICML workshop]
 
+### Formalism for AL
+
+- A Markovian Formalism for Active Querying [2023]
+
 # To be classified
 
 - Active Learning with Label Comparisons [2022]