|
|
|
| Research Advances on Adaptive Perception and Learning in Changing Environment |
| ZHANG Xuyao1,2, YUAN Xiaotong3, LIU Chenglin1,2 |
1. State Key Laboratory of Multimodal Artificial Intelligence Systems, Institute of Automation, Chinese Academy of Sciences, Beijing 100190;
2. School of Artificial Intelligence, University of Chinese Academy of Sciences, Beijing 100049;
3. School of Computer Science, Nanjing University of Information Science and Technology, Nanjing 210044 |
|
|
|
|
Abstract The research on artificial intelligence is gradually extended to open environment from closed environment. There are various changing factors in open environment leading to evident performance degradation of the traditional models and learning algorithms based on closed set assumption and independently and identically distributed assumption. Therefore, adaptive perception and learning in changing environments is a frontier topic in the field of artificial intelligence. The latest advances are introduced from three aspects. For category changing, research issues of open set recognition and out-of-distribution detection, new categories discovery and class-incremental learning are introduced. For data distribution changing, issues of domain adaptation, domain generalization and test-time adaptation are introduced. For data quality changing, issues of weakly supervised learning and label noise learning are introduced. Finally, future research trends are analyzed and discussed.
|
|
Received: 12 October 2023
|
|
|
| Fund:Supported by National Key Research and Development Program of China(No.2018AAA0100400); National Natural Science Foundation of China(No.62222609); National Natural Science Foundation of China(No.62076236) |
|
Corresponding Authors:
ZHANG Xuyao, Ph.D., professor. His research interests include pattern recognition, machine learning and deep learning.
|
About author:: YUAN Xiaotong, Ph.D., professor. His research interests include machine learning, stochastic optimization and computer vision.
LIU Chenglin, Ph.D., professor. His research interests include pattern recognition, machine learning, and document analysis and recognition. |
|
|
|
[1] JAIN A K, DUIN R P W, MAO J C. Statistical Pattern Recognition: A Review. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(1): 4-37.
[2] GOODFELLOW I, BENGIO Y, COURVILLE A. Deep Learning. Cambridge, USA: MIT Press, 2016.
[3] ZHOU Z H.Open-Environment Machine Learning. National Science Review, 2022, 9(8). DOI: 10.1093/nsr/nwac123.
[4] ZHANG X Y, LIU C L, SUEN C Y.Towards Robust Pattern Re-cognition: A Review. Proceedings of the IEEE, 2020, 108(6): 894-922.
[5] ZHANG X Y, XIE G S, LI X Y, et al.A Survey on Learning to Reject. Proceedings of the IEEE, 2023, 111(2): 185-215.
[6] SALEHI M, MIRZAEI H, HENDRYCKS D, et al. A Unified Survey on Anomaly, Novelty, Open-Set, and Out-of-Distribution Detection: Solutions and Future Challenges[C/OL].[2023-09-25]. https://arxiv.org/pdf/2110.14051.pdf.
[7] SUN Y Y, MING Y F, ZHU X J, et al.Out-of-Distribution Detection with Deep Nearest Neighbors // Proc of the 39th International Conference on Machine Learning. San Diego, USA: JMLR, 2022: 20827-20840.
[8] WANG H Q, LI Z Z, FENG L T, et al.VIM: Out-of-Distribution with Virtual-Logit Matching // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 4911-4920.
[9] WEI H X, XIE R C Z, CHENG H, et al. Mitigating Neural Network over Confidence with Logit Normalization // Proc of the 39th International Conference on Machine Learning. San Diego, USA: JMLR, 2022: 23631-23644.
[10] FANG Z, LI Y X, LU J, et al. Is Out-of-Distribution Detection Learnable?[C/OL]. [2023-09-25]. https://openreview.net/pdf?id=sde_7ZzGXOE.
[11] MING Y F, FAN Y, LI Y X.POEM: Out-of-Distribution Detection with Posterior Sampling // Proc of the 39th International Conference on Machine Learning. San Diego, USA: JMLR, 2022: 15650-15665.
[12] KATZ-SAMUELS J, NAKHLEH J B, NOWAK R, et al.Training OOD Detectors in Their Natural Habitats // Proc of the 39th International Conference on Machine Learning. San Diego, USA: JMLR, 2022: 10848-10865.
[13] ZHU F, CHENG Z, ZHANG X Y, et al.OpenMix: Exploring Outlier Samples for Misclassification Detection // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2023: 12074-12083.
[14] MING Y F, CAI Z Y, GU J X, et al. Delving into Out-of-Distribution Detection with Vision-Language Representations[C/OL].[2023-09-25]. https://arxiv.org/abs/2211.13445.
[15] VAZE S, HAN K, VEDALDI A, et al. Open-Set Recognition: A Good Closed-Set Classifier Is All You Need[C/OL].[2023-09-25]. https://arxiv.org/pdf/2110.06207.pdf.
[16] YANG H M, ZHANG X Y, YIN F, et al.Convolutional Prototype Network for Open Set Recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(5): 2358-2370.
[17] HUANG H Z, WANG Y, HU Q H, et al.Class-Specific Semantic Reconstruction for Open Set Recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(4): 4214-4228.
[18] HAN K, REBUFFI S, EHRHARDT S, et al.AutoNovel: Automatically Discovering and Learning Novel Visual Categories. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 44(10): 6767-6781.
[19] ZHONG Z, FINI E, ROY S, et al.Neighborhood Contrastive Lear-ning for Novel Class Discovery // Proc of the IEEE/CVF Confe-rence on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 10862-10870.
[20] FINI E, SANGINETO E, LATHUILIERE S, et al.A Unified Objective for Novel Class Discovery // Proc of the IEEE/CVF Confe-rence on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 9264-9272.
[21] LI W B, FAN Z C, HUO J, et al.Modeling Inter-Class and Intra-Class Constraints in Novel Class Discovery // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2023: 3449-3458.
[22] VAZE S, HAN K, VEDALDI A, et al.Generalized Category Discovery // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 7482-7491
[23] PU N, ZHONG Z, SEBE N.Dynamic Conceptional Contrastive Lear-ning for Generalized Category Discovery // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2023: 7579-7588.
[24] ZHANG S, KHAN S, SHEN Z Q, et al.PromptCAl: Contrastive Affinity Learning via Auxiliary Prompts for Generalized Novel Category Discovery // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2023: 3479-3488.
[25] JOSEPH K J, PAUL S, AGGARWAL G, et al.Novel Class Discovery without Forgetting // Proc of the European Conference on Computer Vision. Berlin, Germany: Springer, 2022: 570-586.
[26] ROY S, LIU M X, ZHONG Z, et al.Class-Incremental Novel Class Discovery // Proc of the European Conference on Computer Vision. Berlin, Germany: Springer, 2022: 317-333.
[27] ZHAO B C, AODHA O M.Incremental Generalized Category Discovery[C/OL]. [2023-09-25].https://arxiv.org/pdf/2304.14310.pdf.
[28] ZHU F, ZHANG X Y, WANG C, et al.Prototype Augmentation and Self-Supervision for Incremental Learning // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 5867-5876.
[29] ZHU F, CHENG Z, ZHANG X Y, et al. Class-Incremental Lear-ning via Dual Augmentation[C/OL].[2023-09-25]. https://proceedings.neurips.cc/paper/2021/file/77ee3bc58ce560b86c2b59363281e914-Paper.pdf.
[30] ZHU K, ZHAI W, CAO Y, et al.Self-Sustaining Representation Expansion for Non-Exemplar Class-Incremental Learning // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 9286-9295.
[31] YAN S P, XIE J W, HE X M.DER: Dynamically Expandable Representation for Class Incremental Learning // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 3013-3022.
[32] DOUILLARD A, RAME A, COUAIRON G, et al.DyTox: Transformers for Continual Learning with Dynamic Token Expansion // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 9275-9285.
[33] WU Z Y, BAEK C, YOU C, et al.Incremental Learning via Rate Reduction // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 1125-1133.
[34] WANG Z F, ZHANG Z Z, LEE C Y, et al.Learning to Prompt for Continual Learning // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 139-149.
[35] LIU X L, HU Y S, CAO X S, et al.Long-Tailed Class Incremental Learning // Proc of the European Conference on Computer Vision. Berlin, Germany: Springer, 2022: 495-512.
[36] KIM G, XIAO C N, KONISHI T, et al.Learnability and Algorithm for Continual Learning // Proc of the 40th International Conference on Machine Learning. San Diego, USA: JMLR, 2023: 16877-16896.
[37] RYMARCZYK D, VAN DER WEIJER J, ZIELINSKI B, et al. IC-ICLE: Interpretable Class Incremental Continual Learning[C/OL].[2023-09-25]. https://arxiv.org/pdf/2303.07811v2.pdf.
[38] ZHU F, ZHANG X Y, WANG R Q, et al.Learning by Seeing More Classes. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(6): 7477-7493.
[39] SHEN M H, BU Y H, WORNELL G W.On Balancing Bias and Variance in Unsupervised Multi-Source-Free Domain Adaptation // Proc of the 39th International Conference on Machine Learning. San Diego, USA: JMLR, 2023: 30976-30991.
[40] CHEN H Y, CHAO W L.Gradual Domain Adaptation without Indexed Intermediate Domains[C/OL]. [2023-09-25].https://arxiv.org/abs/2207.04587.
[41] LI B, WANG Y Z, ZHANG S H, et al.Learning Invariant Representations and Risks for Semi-Supervised Domain Adaptation // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 1104-1113.
[42] SINGH A.CLDA: Contrastive Learning for Semi-Supervised Domain Adaptation[C/OL]. [2023-09-25].https://arxiv.org/pdf/2107.00085v2.pdf.
[43] YANG L Y, WANG Y, GAO M F, et al.Deep Co-training with Task Decomposition for Semi-Supervised Domain Adaptation // Proc of the IEEE/CVF International Conference on Computer Vision. Washington, USA: IEEE, 2021: 8886-8896.
[44] XIE M, LI Y X, WANG Y B, et al.Learning Distinctive Margin Toward Active Domain Adaptation // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 7983-7992.
[45] PRABHU V, CHANDRASEKARAN A, SAENKO K, et al.Active Domain Adaptation via Clustering Uncertainty-Weighted Embe-ddings // Proc of the IEEE/CVF International Conference on Computer Vision. Washington, USA: IEEE, 2021: 8485-8494.
[46] XIE M X, LI S, ZHANG R, et al. Dirichlet-Based Uncertainty Calibration for Active Domain Adaptation[C/OL].[2023-09-25]. https://arxiv.org/abs/2302.13824.
[47] HUANG X, BELONGIE S.Arbitrary Style Transfer in Real-Time with Adaptive Instance Normalization // Proc of the IEEE International Conference on Computer Vision. Washington, USA: IEEE, 2017: 1510-1519.
[48] ZHOU K Y, YANG Y X, QIAO Y, et al. Domain Generalization with Mixstyle[C/OL].[2023-09-25]. https://arxiv.org/pdf/2104.02008.pdf.
[49] HUANG J X, GUAN D Y, XIAO A, et al.FSDR: Frequency Space Domain Randomization for Domain Generalization // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 6887-6898.
[50] SHANKAR S, PIRATLA V, CHAKRABARTI S, et al. Generalizing across Domains via Cross-Gradient Training[C/OL].[2023-09-25]. https://arxiv.org/pdf/1804.10745.pdf.
[51] ZHANG J, ZHANG X Y, WANG C, et al.Deep Representation Learning for Domain Generalization with Information Bottleneck Principle. Pattern Recognition, 2023, 143. DOI: 10.1016/j.patcog.2023.109737.
[52] BAI H Y, SUN R, HONG L Q, et al.DecAug: Out-of-Distribution Generalization via Decomposed Feature Representation and Semantic Augmentation // Proc of the AAAI Conference on Artificial Intelligence. Palo Alto, USA: AAAI Press, 2021: 6705-6713.
[53] NAM H, LEE H, PARK J, et al.Reducing Domain Gap by Reducing Style Bias // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 8686-8695.
[54] ARJOVSKY M, BOTTOU L, GULRAJANI I, et al. Invariant Risk Minimization[C/OL].[2023-09-25]. https://arxiv.org/pdf/1907.02893.pdf.
[55] LIN Y, DONG H Z, WANG H, et al.Bayesian Invariant Risk Minimization // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 16000-16009.
[56] SHI Y G, SEELY J, TORR P H S, et al. Gradient Matching for Domain Generalization[C/OL].[2023-09-25]. https://arxiv.org/pdf/2104.09937.pdf.
[57] BUCCI S, D'INNOCENTE A, LIAO Y J, et al. Self-Supervised Learning Across Domains. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(9): 5516-5528.
[58] SHAH H, TAMULY K, RAGHUNATHAN A, et al. The Pitfalls of Simplicity Bias in Neural Networks[C/OL].[2023-09-25]. https://arxiv.org/pdf/2006.07710.pdf.
[59] GEIRHOS R, JACOBSEN J H, MICHAELIS C, et al.Shortcut Learning in Deep Neural Networks. Nature Machine Intelligence, 2020, 2(11): 665-673.
[60] TENEY D, ABBASNEJAD E, VAN DER LUCEY S, et al. Evading the Simplicity Bias: Training a Diverse Set of Models Disco-vers Solutions with Superior OOD Generalization // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 16740-16751.
[61] WANG D Q, SHELHAMER E, LIU S T, et al. Tent: Fully Test-Time Adaptation by Entropy Minimization[C/OL].[2023-09-25]. https://arxiv.org/abs/2006.10726.
[62] CHOI S, YANG S H, CHOI S, et al.Improving Test-Time Adaptation via Shift-Agnostic Weight Regularization and Nearest Source Prototypes // Proc of the European Conference on Computer Vision. Berlin, Germany: Springer, 2022: 440-458.
[63] WANG Q, FINK O, VAN GOOL L, et al.Continual Test-Time Domain Adaptation // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 7191-7201.
[64] NIU S C, WU J X, ZHANG Y F, et al. Efficient Test-Time Model Adaptation without Forgetting[C/OL].[2023-09-25]. https://arxiv.org/pdf/2204.02610.pdf.
[65] SONG J H, LEE J, KWEON I S, et al.EcoTTa: Memory-Efficient Continual Test-Time Adaptation via Self-Distilled Regularization // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2023: 11920-11929.
[66] ZHANG M, LEVINE S, FINN C.MEMO: Test Time Robustness via Adaptation and Augmentation[C/OL]. [2023-09-25].https://arxiv.org/pdf/2110.09506v2.pdf.
[67] LIM H, KIM B, CHOO J, et al. TTN: A Domain-Shift Aware Batch Normalization in Test-Time Adaptation[C/OL].[2023-09-25]. https://arxiv.org/abs/2302.05155.
[68] NIU S C, WU J X, ZHANG Y F, et al. Towards Stable Test-Time Adaptation in Dynamic Wild World[C/OL].[2023-09-25]. https://arxiv.org/pdf/2302.12400.pdf.
[69] WANG Y Q, YAO Q M, KWOK J T, et al.Generalizing from a Few Examples: A Survey on Few-Shot Learning. ACM Computing Surveys, 2020, 53(3). DOI: 10.1145/3386252.
[70] LU J, GONG P H, YE J P, et al.A Survey on Machine Learning from Few Samples. Pattern Recognition, 2023, 139. DOI: 10.1016/j.patcog.2023.109480.
[71] FINN C, ABBEEL P, LEVINE S.Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks // Proc of the 34th International Conference on Machine Learning. San Diego, USA: JMLR, 2017: 1126-1135.
[72] JAMAL M A, QI G J.Task Agnostic Meta-Learning for Few-Shot Learning // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2019: 11711-11719.
[73] NAJDENKOSKA I, ZHEN X T, WORRING M.Meta Learning to Bridge Vision and Language Models for Multimodal Few-Shot Learning[C/OL]. [2023-09-25].https://openreview.net/pdf?id=3oWo92cQyxL.
[74] KOCH G, ZEMEL R, SALAKHUTDINOV R.Siamese Neural Networks for One-Shot Image Recognition[C/OL]. [2023-09-25].https://www.cs.cmu.edu/~rsalakhu/papers/oneshot1.pdf.
[75] AFRASIYABI A, LAROCHELLE H, LALONDE J, et al.Matching Feature Sets for Few-Shot Image Classification // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2022: 9004-9014.
[76] WEI X S, XU H Y, ZHANG F, et al. An Embarrassingly Simple Approach to Semi-Supervised Few-Shot Learning[C/OL].[2023-09-25]. https://arxiv.org/pdf/2209.13777.pdf.
[77] ZHOU Z H.A Brief Introduction to Weakly Supervised Learning. National Science Review, 2018, 5(1): 44-53.
[78] ZHOU Y J, SONG X C, ZHANG Y R, et al.Feature Encoding with Autoencoders for Weakly Supervised Anomaly Detection. IEEE Transactions on Neural Networks and Learning Systems, 2022, 33(6): 2454-2465.
[79] SHI Y F, XU X, XI J H, et al.Learning to Detect 3D Symmetry from Single-View RGB-D Images with Weak Supervision. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(4): 4882-4896.
[80] WANG W, GAO J Y, XU C S.Weakly-Supervised Video Object Grounding via Causal Intervention. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(3): 3933-3948.
[81] FAN J S, ZHANG Z X.Memory-Based Cross-Image Contexts for Weakly Supervised Semantic Segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2023, 45(5): 6006-6020.
[82] LI Y F, GUO L Z, ZHOU Z H.Towards Safe Weakly Supervised Learning. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 43(1): 334-346.
[83] GONG C, YANG J, YOU J, et al.Centroid Estimation with Gua-ranteed Efficiency: A General Framework for Weakly Supervised Learning. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2022, 44(6): 2841-2855.
[84] JIANG L, ZHOU Z Y, LEUNG T, et al.MentorNet: Learning Data-Driven Curriculum for Very Deep Neural Networks on Corrupted Labels // Proc of the 35th International Conference on Machine Learning. San Diego, USA: JMLR, 2018: 2304-2313.
[85] HAN B, YAO Q M, YU X R, et al.Co-teaching: Robust Training of Deep Neural Networks with Extremely Noisy Labels // Proc of the 32nd International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2018: 8536-8546.
[86] YU X R, HAN B, YAO J C, et al.How Does Disagreement Help Generalization Against Label Corruption? // Proc of the 36th International Conference on Machine Learning. San Diego, USA: JMLR, 2019: 7164-7173.
[87] XIA X R, LIU T L, HAN B, et al. Sample Selection with Uncertainty of Losses for Learning with Noisy Labels[C/OL].[2023-09-25]. https://arxiv.org/pdf/2106.00445.pdf.
[88] WEI J S, LIU H Y, LIU T L, et al. To Smooth or Not? When Label Smoothing Meets Noisy Labels[C/OL]. [2023-09-25]. https://arxiv.org/pdf/2106.04149.pdf
[89] LI Y F, HAN H, SHAN S G, et al.DISC: Learning from Noisy Labels via Dynamic Instance-Specific Selection and Correction // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2023: 24070-24079.
[90] LI J N, SOCHER R, HOI S C H. DivideMix: Learning with Noisy Labels as Semi-Supervised Learning[C/OL].[2023-09-25]. https://arxiv.org/abs/2002.07394.
[91] NISHI K, DING Y, RICH A, et al.Augmentation Strategies for Learning with Noisy Labels // Proc of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Washington, USA: IEEE, 2021: 8018-8027.
[92] BAI Y B, YANG E, HAN B, et al. Understanding and Improving Early Stopping for Learning with Noisy Labels[C/OL].[2023-09-25]. https://arxiv.org/pdf/2106.15853v2.pdf.
[93] WEI Q, SUN H L, LU X K, et al.Self-Filtering: A Noise-Aware Sample Selection for Label Noise with Confidence Penalization // Proc of the European Conference on Computer Vision. Berlin, Germany: Springer, 2022: 516-532.
[94] ZHANG Z L, SABUNCU M R.Generalized Cross Entropy Loss for Training Deep Neural Networks with Noisy Labels // Proc of the 32nd International Conference on Neural Information Processing Systems. Cambridge, USA: MIT Press, 2018: 8792-8802.
[95] LI X C, XIA X B, ZHU F, et al.Dynamics-Aware Loss for Lear-ning Withlabel Noise. Pattern Recognition, 2023, 144. DOI: 10.1016/j.patcog.2023.109835.
[96] WANG H C, FU T F, DU Y Q, et al.Scientific Discovery in the Age of Artificial Intelligence. Nature, 2023, 620: 47-60. |
|
|
|
2023, Vol. 36 
