Curriculum learning
The DOI matches its Crossref record; 35 of the 40 references it cites were found in a registry or an index.
Abstract
Humans and animals learn much better when the examples are not randomly presented but organized in a meaningful order which illustrates gradually more concepts, and gradually more complex ones. Here, we formalize such training strategies in the context of machine learning, and call them "curriculum learning". In the context of recent research studying the difficulty of training in the presence of non-convex training criteria (for deep deterministic and stochastic neural networks), we explore curriculum learning in various set-ups. The experiments show that significant improvements in generalization can be achieved. We hypothesize that curriculum learning has both an effect on the speed of convergence of the training process to a minimum and, in the case of non-convex criteria, on the quality of the local minima obtained: curriculum learning can be seen as a particular form of continuation method (a general strategy for global optimization of non-convex functions).
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“Moreover, the lack of acoustical signals could even account for spatial disorientation 91,92.”
From The impact of multisensory integration and perceptual load in virtual reality settings on performance, workload and presence (2021) · cites this paper as [92]
In the citing paper
“This suggests that curriculum-style learning (Bengio et al., 2009), where simpler or archetypal problems are presented first to build a repository of valid heuristics, may potentially bootstrap performance.”
From Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory (2025) · cites this paper as Bengio et al., 2009
Citations
Cites
Reducing the dimensionality of data with neural networks
Hinton, G. E., & Salakhutdinov, R. R. (2006) · Science · doi:10.1126/science.1127647 · Cited by 20,928
A fast learning algorithm for deep belief nets
Hinton, G. E., Osindero, S., & Teh, Y.-W. (2006) · Neural Computation · doi:10.1162/neco.2006.18.7.1527 · Cited by 16,336
Learning Deep Architectures for AI
Bengio, Y. (2009) · Foundations and Trends® in Machine Learning · doi:10.1561/2200000006 · Cited by 9,239
Self-citationExtracting and composing robust features with denoising autoencoders
Vincent, P., Larochelle, H., Bengio, Y., & Manzagol, P.-A. (2008) · Proceedings of the 25th international conference on Machine learning - ICML '08 · doi:10.1145/1390156.1390294 · Cited by 7,436
Self-citationA unified architecture for natural language processing: deep neural networks with multitask learning
Collobert, R., & Weston, J. (2008) · Proceedings of the 25th international conference on Machine learning - ICML '08 · doi:10.1145/1390156.1390177 · Cited by 5,058 · cited 6×
“In Collobert and Weston (2008), the main objective is to learn an embedding for the words as a side-e.ect of learning to compute this score.”
BackgroundMethodInfluentialSelf-citationLearning Deep Architectures for AI
Bengio, Y. (2009) · now publishers, Inc. eBooks · doi:10.1561/9781601982957 · Cited by 4,941
Self-citationGreedy layer-wise training of deep networks
Bengio, Y., Lamblin, P., Popovici, D., & Larochelle, H. (2007) · Advances in Neural Information Processing Systems 19 · doi:10.7551/mitpress/7503.003.0024 · Cited by 4,841
Self-citationA Neural Probabilistic Language Model
Bengio, Y., Ducharme, R., Vincent, P., Jauvin, C., Jaz, K., Hofmann, T., Poggio, T., & Shawe-Taylor, J. S. (2000) · Cited by 3,479
Self-citationLearning and development in neural networks: the importance of starting small
Elman, J. L. (1993) · Cognition · doi:10.1016/0010-0277(93)90058-4 · Cited by 1,988 · cited 8×
“The idea of training a learning machine with a curriculum can be traced back at least to Elman (1993).”
BackgroundMethodInfluentialRestricted Boltzmann machines for collaborative filtering
Salakhutdinov, R., Mnih, A., & Hinton, G. (2007) · Proceedings of the 24th international conference on Machine learning · doi:10.1145/1273496.1273596 · Cited by 1,885
Research works printed but not linked: 3
- 10.5555/89603 · doi:10.5555/89603
- 10.5555/866740 · doi:10.5555/866740
- 10.5555/525571 · doi:10.5555/525571
Cited by
Items 4,701–4,800 of 5,239
Jin, Z., Li, J., & Zhang, R. (2025) · International Conference on Energy and Electrical Power Systems (ICEEPS) · doi:10.1109/iceeps66790.2025.11239705
“Such an approach is conceptually inspired by dynamic curriculum learning strategies, where task difficulty is adjusted in response to model progress or uncertainty [16] .”
Libra: Large Chinese-based Safeguard for AI Content
Chen, Z., Yu, H., Wu, X. Y., Liu, D., & Hu, S. (2025) · Lecture notes in computer science · doi:10.1007/978-981-95-3343-5_44
Tsourveloudis, C., & Doitsidis, L. (2026) · Journal of Intelligent & Robotic Systems · doi:10.1007/s10846-026-02408-7 · Cited by 3
“Curriculum learning [15] has proven effective for accelerating learning and improving final performance, when tasks are presented in order of increasing difficulty.”
Faro: A Scalable and Reliable Outage Detection Algorithm for IoT Mobile Virtual Network Aggregators
Milani, M., Vomhoff, V., Bega, D., Gramaglia, M., Geißler, S., Karsai, C., & Serrano, P. (2025) · Proceedings of the ACM on Networking · doi:10.1145/3768982
PLATO-JDS: Enhancing Japanese Dialogue Systems Through Topic-Switch Adaptation
Zeng, D., Wu, J., Wang, Y., Matsumoto, K., Hattori, G., & Ikeda, K. (2025) · New Generation Computing · doi:10.1007/s00354-025-00307-3
Entropy-Driven Curriculum for Multi-Task Training in Human Mobility Prediction
Fang, T., Luo, X., & Werner, M. (2025) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata66926.2025.11401749 · Cited by 2
“Third, this approach naturally implements a form of regularization, as the model must maintain performance on simple examples while adapting to complex ones, thereby preventing overfitting to edge cases [21].”
Training People to Reward Robots
Sun, E., Zhu, Y., & Howard, M. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11245938
“2) Scaffolding Training: The approach of progressively increasing task complexity, often referred to as scaffolding training or curriculum learning , is particularly effective in educating novices [14], [15].”
Abouelazm, A., Ratz, J., Schörner, P., & Zöllner, J. M. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11246550
“To develop driving skills more comprehensively, some works adopt curriculum learning [11], a technique inspired by education that trains agents on simpler tasks before progressively advancing to more complex ones.”
TERL: Large-Scale Multi-Target Encirclement Using Transformer-Enhanced Reinforcement Learning
Zhang, H., Zhao, G.-X., & Ren, X.-Q. (2025) · IEEE/RJS International Conference on Intelligent RObots and Systems · doi:10.1109/iros60139.2025.11246832 · Cited by 5
“Curriculum learning improves model performance by progressively introducing training examples from simple to complex, facilitating convergence and generalization [34].”
Learning-Based Quadruped Robot Framework for Locomotion on Dynamic Rigid Platforms
Huang, K., Feng, H., Meng, W., Wei, T., & Hu, T. (2025) · IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) · doi:10.1109/iros60139.2025.11247628 · Cited by 1
“Curriculum learning [44], [45], [46] is also applied to improve training efficiency.”
Wang, Y. (2025) · Scientific Reports · doi:10.1038/s41598-025-26405-2 · Cited by 1
“The optimization of training strategies incorporates curriculum learning principles that progressively introduce more complex combat scenarios as the network develops proficiency with fundamental movements 43 .”
Curriculum-Structured Fine-Tuning for Citation-Faithful QA under the Thai Land and Buildings Tax ActPreprint
Tantapong, N., Arsaibun, S., Kantavat, P., Meknavin, S., & Kijsirikul, B. (2025) · Research Square · doi:10.21203/rs.3.rs-8142832/v1
An intelligent framework for dynamic modeling of therapeutic response using clinical compliance data
Wang, X.-Y., & Xie, H. (2025) · Frontiers in Pharmacology · doi:10.3389/fphar.2025.1631599
Extending Learning-by-Asking to Real-World Visual Question Answering
Netra, S., Vijarnia, A., Bhambri, S., & Malhotra, P. (2025) · IEEE International Conference on Contemporary Computing and Communications (InC4) · doi:10.1109/inc465408.2025.11256379
“2) Curriculum learning techniques to optimize question selection dynamically [9].”
InfluentialThree Birds with One Stone: Improving Performance, Convergence, and System Throughput with NEST
Huo, Y., Quiroga, D., Kyrillidis, A., & Patel, T. (2025) · ACM SIGMETRICS Performance Evaluation Review · doi:10.1145/3771580 · Cited by 2
Lavín-Pallero, E., & Ruiz-García, M. (2025) · Journal of Statistical Mechanics Theory and Experiment · doi:10.1088/1742-5468/ae1b86 · Cited by 3
“In deep learning, there are other methods that change the loss function during learning, a paradigmatic example is curriculum learning [46], where the training examples are not randomly presented but are organized in a meaningful order.”
Robust Visual Transformers for Medical Image Classification
Montrezol, J. P., Oliveira, H. S., Araújo, J., et al. (2025) · doi:10.1109/embc58623.2025.11253903
“ProCAN leverages a progressive growing strategy to learn hierarchical features by gradually increasing image resolution in a Curriculum Learning [15] manner, the stem of this architecture consists of a modified attention mechanism for attention capture in more significant feature channels.”
Schmale, S., Hempel, P., & Spicher, N. (2025) · Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC) · doi:10.1109/embc58623.2025.11252863 · Cited by 1
“It has been shown in other classification problems, that robustness as well as generalization capabilities can be improved by an adjusted training order, i.e. starting with easy-to-learn and ending with difficult-to-learn samples with the concept being titled ”Curriculum Learning” (CL) [19]–[21].”
Difficulty-Aware Adaptive Augmentation Scheduling for SAR Semi-Supervised Learning
Choung, H., & Kim, S. (2025) · IEEE International Conference on Consumer Electronics-Asia (ICCE-Asia) · doi:10.1109/icce-asia67487.2025.11263570
“This design reduces reliance on static augmentation policies, enhances data utilization, and is distinguished by its ability to estimate sample difficulty online during the SSL process and reflect it in the augmentation policy [7], [15], [16].”
Enhancing Test-Time Training for Corruption Robustness in Image Classification
Park, M., Bae, J., & Shin, J. (2025) · IEEE International Conference on Consumer Electronics-Asia (ICCE-Asia) · doi:10.1109/icce-asia67487.2025.11263571
“Moreover, adapting to incoming samples in a random order is rarely optimal, since the order and composition of the test stream can shape convergence and stability [9].”
Mumu, F. C., Iqbal, K. I. B., & Du, L. (2025) · ACM International Conference on Multimedia in Asia (MMAsia) · doi:10.1145/3743093.3771061
Yoon, Y., Kang, B., Choi, J.-W., & Yun, W. (2025) · Journal of the Korea Institute of Military Science and Technology · doi:10.9766/kimst.2025.28.6.699
“수동 방식은 설계자의 주관이 개입되어 재현성 이 떨어지는 한계가 있으며 [9] , 완전 자동화 방식은 객 관성을 확보할 수 있지만 알고리즘이 복잡하고 안전 이 중요한 국방 분야에 적용하기에는 통제 가능성이 떨어진다는 우려가 있다 [8] .”
Industrial anomaly detection via curriculum-based deep learning
Dube, D. C., & Akar, M. (2025) · Applied Intelligence · doi:10.1007/s10489-025-07032-w
“Bengio et al. [15] introduced curriculum learning (CL), which is a strategy for training ML models by starting with “easy” examples and gradually introducing “difficult” ones.”
From Sparse to Dense: Toddler-Inspired Reward Transition in Goal-Oriented Reinforcement Learning
Park, J., Yang, H., Lee, M. W., Choi, W.-S., Lee, M., & Zhang, B. (2025) · IEEE Transactions on Cognitive and Developmental Systems · doi:10.1109/tcds.2025.3642038 · Cited by 5
“:3) is characterized by a unique reward function R i , representing different task difficulties [32], [33].”
Narkedimilli, S., Kumar, C. P., & Ramachandra, R. (2025) · Lecture notes in computer science · doi:10.1007/978-3-032-13714-2_18
Federated Curriculum Learning for Privacy-Preserving Personalization in Edge Environments
Mehta, A. D. (2025) · Darpan International Research Analysis · doi:10.36676/dira.v13.i4.182
ACE-PPO and MDE-IDS: a novel approach for efficiency and security in 5G optical networks
Kulshreshtha, P., & GARG, A. K. (2025) · Journal of Optical Communications · doi:10.1515/joc-2025-0398
Ning, A., Yang, L., & Chern, G.-W. (2025) · Physical review. E · doi:10.1103/7dr5-68ky · Cited by 1
Recurrent Neural Networks Meet Context-Free Grammar: Two Birds with One Stone
Guan, H., Chaudhary, U. K., Xu, Y., Ning, L., Zhang, L., & Shen, X. (2026) · ACM Transactions on Knowledge Discovery from Data · doi:10.1145/3785464
“Curriculum Learning [5] is a training strategy inspired by the way humans learn, where models are trained on tasks or data in a meaningful order, starting with simpler examples and gradually progressing to more complex ones.”
Ullmo, M., Aghanim, N., Decelle, A., & Aragón-Calvo, M. A. (2025) · Astronomy and Astrophysics · doi:10.1051/0004-6361/202449845 · Cited by 4
DeepVoid: A Deep Learning Void Detector
Kumagai, S., Vogeley, M. S., Aragón-Calvo, M. A., Douglass, K. A., BenZvi, S., & Neyrinck, M. C. (2026) · The Astrophysical Journal · doi:10.3847/1538-4357/ae2c80 · Cited by 3
“Curriculum learning (CL) is a training strategy used in machine learning that involves beginning with ‘easier’ tasks and progressively increasing the tasks’ difficulty (Bengio et al. 2009; Soviany et al. 2021).”
Federated deep convolutional neural network architecture for renal cancer classification
Yu, X., Lin, Z., Zhou, C., Wang, H., & Wang, M. (2025) · Discover Computing · doi:10.1007/s10791-025-09870-1 · Cited by 2
“Curriculum Learning (CL) was first introduced by Bengio [34], inspired by the human learning process.”
Halliday, R. B. (2005) · INFM-OAR (INFN Catania) · doi:10.5281/zenodo.170360
CULC-Net: A Recipe for Tailored Creative Selection in Online Advertising
Zhang, B., Sang, L., Wang, H., Wang, W., Chen, W., Peng, C., Lin, Z., Shao, J., He, J., Wang, H., et al. (2025) · Lecture notes in computer science · doi:10.1007/978-3-032-06096-9_30
LLM-SYM: Integrating Symbolic Methods and Large Language Models for Automated Theorem Proving
Wu, Y., Huang, Y., & Shi, J. (2025) · Lecture notes in computer science · doi:10.1007/978-981-95-4213-0_3
Leveraging Self-Paced Learning for Software Vulnerability Detection
Cheng, Z., Yang, Y., Zhang, H., Yang, L., Hu, J., Xu, J., Liu, B., & Shen, H. (2026) · ACM Transactions on Software Engineering and Methodology · doi:10.1145/3844141 · Cited by 1
“Self-paced learning is a machine learning strategy that simulates the human autonomous learning process [1], in which students adjust their learning pace and curriculum based on their knowledge mastery.”
Towards Frame Rate Insensitive Video Object Segmentation
Chong, W., Huang, K., Shi, M., Li, J., Yang, H., & Zhang, D. (2025) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn64981.2025.11227650 · Cited by 1
“We propose the SimInfer base on curriculum learning [1].”
Robust Monocular Visual Odometry via Dual-Paradigm Curriculum Learning
Lahiany, A., & Gal, O. (2025) · IEEE Robotics and Automation Letters · doi:10.1109/lra.2025.3635447
“Curriculumlearning(CL)[13]structuresdataexposureto improve robustness, evolving from manual design to self-paced methods [14] and automatic discovery [15].”
VDC-Agent: When Video Detailed Captioners Evolve Themselves via Agentic Self-Reflection
Wang, Q., Gao, X., He, Y., Han, J., Li, J., Dong, S., Ma, Z., & Gong, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37592-6_30 · Cited by 5
Multi-talker audio–visual speech recognition towards diverse scenarios
Lin, Y., Jin, T., Cheng, X., Zhao, Z., & Wu, F. (2025) · Frontiers of Information Technology & Electronic Engineering · doi:10.1631/fitee.2500411
“Curriculum learning, first proposed by Bengio et al. (2009), is the idea of arranging training data in a meaningful order, similar to human learning.”
Chain-of-Thought Curriculum Distillation: Teaching Smaller Models to Reason Step-by-Step
Yu, X., Huo, Y., Cai, L., & Luo, X. (2025) · International Symposium on Artificial Intelligence for Education (ISAIE) · doi:10.1145/3775073.3775200 · Cited by 2
Differentiable Physics Training Method for Robot Motion Planning
Leja, L., Strazds, G. V., Chronis, C., Varlamis, I., & Freivalads, K. (2025) · IFAC-PapersOnLine · doi:10.1016/j.ifacol.2025.12.106
Expert Guided Diffusion for Synthetic Aperture Radar Despeckling
Lin, P.-C., Ushijima-Mwesigwa, H., & Au, W. Y. (2025) · ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL) · doi:10.1145/3748636.3763221
“Additionally, we incorporate curriculum learning [1], a training technique that organizes data by increasing difficulty.”
Towards Superhuman Imitation Learning for Sequential Head-and-Neck Cancer Treatment DecisionsPreprint
Corna, F., Zhang, X., Canahuate, G., et al. (2025) · medRxiv · doi:10.64898/2025.12.11.25342119
Fan, R., Chen, D., Wang, C., Wang, T., Song, J., Chen, Y., Tong, T., & Zhang, X. (2025) · The Visual Computer · doi:10.1007/s00371-025-04292-y
“Curriculum learning [28] is a training strategy that improves a model’s efficiency and generalization by starting with simple data and gradually introducing more complex data, mimicking the "easy-to-difficult" learning process in humans.”
Post Hoc Error Correction for Missing Classes in Deep Neural Networks
Lebedev, A. A., Kazantsev, V., & Stasenko, S. V. (2025) · Technologies · doi:10.3390/technologies14010008
Qin, G., Li, B., Li, R., Wang, Y., Zhao, H. B., & Fan, X. (2025) · Photonics · doi:10.3390/photonics13010010
Beyond Random Masking: A Dual-Stream Approach for Rotation-Invariant Point Cloud Masked Autoencoders
Yin, X., Zhang, D., Feng, Y., Mao, S., Yu, J., & Cai, W. (2025) · International Conference on Digital Image Computing: Techniques and Applications (DICTA) · doi:10.1109/dicta68720.2025.11302501 · Cited by 1
“The dynamic weighting creates a curriculum learning [6] approach with three distinct phases.”
InfluentialUsing LLM to Design Reward Function for Bipedal Walker-v3
Dong, S. (2025) · International Conference on Artificial Intelligence and Intelligent Information Processing (AIIIP) · doi:10.1145/3778534.3778593
“The reason the final version could perfectly achieve human-like walking is likely that the LLM implemented a curriculum learning [2] approach in the reward function.”
DreamerRL: Um Framework de RL para o Desenvolvimento Autônomo em Robótica Humanoide
Correia, A. D. S., Costa, P. D. P., & Colombini, E. L. (2025) · doi:10.5753/sbrlars_estendido.2025.248267
“Mesmo com o avanc¸o de t´ecnicas como transfer learning [Shah and Kumar 2021], reinforcement learning [Sun et al. 2025] e curriculum learning [Bengio et al. 2009], a aquisic¸˜ao de habilidades ´uteis e generaliz´aveis ainda requer consider´avel intervenc¸˜ao humana, seja na engenharia de tarefas,…”
Sparse Mixture-of-Experts LLMs for Efficient ICD Code Prediction in Clinical Outcome Modeling
Dantale, N., Goel, P., Mundra, S., Mundra, A., & Aggarwal, G. (2025) · International Conference on Contemporary Computing · doi:10.1109/ic366947.2025.11290394
“2) Cross-Modal Instruction Tuning: Fine-tuning is performed using a curriculum learning approach [9] with an initial learning rate of 5 × 10 − 5 over 20 epochs.”
A single-station hourly multi-step forecasting model for sea fog based on ResNet and LSTM networks
Sun, J., Qian, Y., Sun, Y., et al. (2026) · Journal of the Meteorological Society of Japan Ser II · doi:10.1007/s44394-025-00004-1
Speed Estimation from Audio Using Vision Transformers
Abdelaziz, S., & Ziani, S. (2026) · Lecture notes in electrical engineering · doi:10.1007/978-981-95-1113-6_45
Curriculum Reinforcement Learning for Robust Low-Resource Chinese Dialect Speech Recognition
Hou, L., Gao, Y., Liu, M., et al. (2026) · Communications in computer and information science · doi:10.1007/978-981-95-5382-2_9
M3F-U-Net: A hybrid deep learning model for precipitation forecast correction
Zhang, T., Wang, X., Li, H., et al. (2026) · Journal of Atmospheric and Solar-Terrestrial Physics · doi:10.1016/j.jastp.2025.106714
Incremental imbalance-aware deep learning framework for multilingual spoken language identification
Tomar, V., Dixit, S., & Sangwan, P. (2026) · Language Resources and Evaluation · doi:10.1007/s10579-025-09877-6
“Incremental learning paradigms, including curriculum learning (Bengio et al., 2009) and elastic weight consolidation (EWC) (Kirkpatrick et al., 2017), provide principled strategies to extend models without catastrophic forgetting.”
Sun, C., Sun, C., Wang, J., et al. (2026) · Engineering Applications of Artificial Intelligence · doi:10.1016/j.engappai.2026.113779
Boosting Active Prompt Learning via Discriminative Self-Training Dual-Curriculum Learning
Tao, S., Liu, J., Zeng, P., Xu, Y., Hu, B., & Zha, Z.-J. (2026) · International Journal of Computer Vision · doi:10.1007/s11263-025-02641-x
“Curriculum learning (CL) [3, 57] is a training strategy inspired by the characteristics of human learning.”
RL-AWB: Deep Reinforcement Learning for Auto White Balance Correction in Low-Light Nighttime Scenes
Lee, Y.-K., Chen, K.-L., Chang, C.-C., & Liu, Y.-L. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37032-7_5 · Cited by 5
“In summary, we make the following contributions: • We develop SGP-LRD (Salient Gray Pixels with Local Reflectance Differences), a nighttime-specific color constancy algorithm that achieves state-of-the-art illumination estimation on public nighttime benchmarks.”
Wang, L. N., Guo, Z., Li, H., & Chen, C. (2025) · IEEE International Conference on Big Data and Artificial Intelligence (BDAI) · doi:10.1109/bdai66031.2025.11325344
“Curriculum Learning (CL), introduced by Bengio [6], optimizes the sequence of training samples to improve performance.”
Akamisaka, A., & Nagao, K. (2026) · Sensors · doi:10.3390/s26020566
Decentralized Reinforcement Learning for Cooperative Multi-Robot Navigation
Zheng, Y., Arvin, F., & Hu, J. (2025) · IEEE International Conference on Parallel and Distributed Systems (ICPADS) · doi:10.1109/icpads67057.2025.11323034
“We further use curriculum learning [31]: training starts with one robot in a 10 × 10 grid; each actor maintains a rolling 100-episode success rate, and once it exceeds 90% we increase difficulty by enlarging the grid and adding one robot, improving stability and generalization.”
Yigin, B. Ö., Saygılı, G., & Spronck, P. H. M. (2026) · IEEE Transactions on Computational Biology and Bioinformatics · doi:10.1109/tcbbio.2026.3654772
“Theconcept of CL is closely related to curriculum learning [13] and self-paced learning [14], both of which emphasize the structured scheduling of training samples to optimize learning efficiency.”
Chen, S., Shi, H., & Zhang, X.-Y. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-5628-1_20
Ahmed, M., Ammar, A., Nasir, Q., Talib, M. W. A., & Mohammed, M. K. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.6105125
Semi-supervised Classification Method for Class-Imbalanced Medical Images Based on Bias Correction
Shen, H., Wang, Z., Wu, Q., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-5634-2_10
Ahmed, A. S., & Mustafa, M. O. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-13921-4_48
Anomaly-Aware Curriculum Learning for Cybersecurity in the Internet of Medical Things
Alouache, Y., Merabti, M., Salem, O., et al. (2025) · IEEE International Conference on E-health Networking, Application & Services (HealthCom) · doi:10.1109/healthcom60686.2025.11342647
Wang, C.-A., Huang, S.-H., Chen, C.-T., & Fang, Y.-T. (2026) · Decision Support Systems · doi:10.1016/j.dss.2026.114624
NeRF-MIR: Toward High-Quality Restoration of Masked Images With Neural Radiance Fields
Huang, X., Zhong, Z., Chen, S., Xu, Y., Guan, J., & Zhou, S. (2026) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2026.3651302 · Cited by 2
“The dynamically weighted loss function follows an intuitive principle inspired by curriculum learning [67], where the model gradually evolves from coarse to fine learning.”
Deep Neural Networks for Block-Level Video Enhancement: Transforming LR AVS to HR HEVC Video
Nawaz, M., Hamza, M. A., Daud, M., et al. (2025) · doi:10.1109/ai-si66213.2025.11341233
“4) Curriculum Learning: Curriculum learning [24] is a training strategy in which training examples are presented to a neural network in a specific order, beginning with easier examples and gradually progressing to more difficult ones.”
Towards Verification of Learning-Based UAS Trajectory Generation in Uncertain Environments
Davies, G., & Sagoo, J. S. (2025) · Safety-Critical Systems Symposium (SSS) · doi:10.65391/r3085
Nishad, K. A. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-14706-6_6
Yuan, P., & Zhao, S. (2026) · Lecture notes in electrical engineering · doi:10.1007/978-981-95-3320-6_16
Hierarchical Semantic Learning for Multi-class Aorta Segmentation
Shi, P. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-14246-7_2
Tuning Distillation to Generate Edge-friendly All-rounder Models
Dey, S., Mukherjee, A., & Pal, A. (2025) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc58881.2025.11343501
“This curriculum setup [38], where the distillation phase acts as a simpler task that eases the transition to more complex SSL objectives, is supported by findings from [34], [15] and helps small models converge faster and generalize better than with standard SSL alone.”
Yin, W., & Zhao, G. (2026) · IEEE Transactions on Neural Networks and Learning Systems · doi:10.1109/tnnls.2026.3655179
“Beginning with the seminal work of [4], curriculum learning has demonstrated its e ff ectiveness in facilitating policy acquisition through progressive task generation with increasing di ffi culty.”
Yuchuan, H., Dang, J., Yang, H., & Peng, B. (2025) · Conference proceedings/Conference proceedings - IEEE International Conference on Systems, Man, and Cybernetics · doi:10.1109/smc58881.2025.11342659
“Additionally, we incorporate a strain-guided curriculum learning (CL) strategy [17] to improve robustness by training the model progressively from easy to hard samples.”
FGit: Fault-Guided Fine-Tuning for Code Generation
Fan, L., Liu, Z., Wang, H., Bao, L., Xia, X., & Li, S. (2025) · IEEE/ACM International Conference on Automated Software Engineering (ASE) · doi:10.1109/ase63991.2025.00114 · Cited by 5
“Drawing inspiration from curriculum learning [11], which involves training models by progressively moving from easier to more challenging samples to build learning capabilities, we propose Fault-Guided Fine-Tuning (FG IT ).”
Tian, Q., Zhang, X., Yang, T., et al. (2025) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm66473.2025.11356600
“Building upon the mT5 model, we introduce a terminology dictionary-constrained decoding mechanism to refine the curriculum learning strategy [11] .”
Zhang, Y., Nailon, W. H., Zhao, T., Wang, G., & Cheng, K. (2025) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm66473.2025.11356529 · Cited by 2
“Motivated by curriculum learning principles that prioritize difficult samples during training [15], [16], we argue that effective spatial calibration should allocate stronger regularization to challenging regions with concentrated prediction errors, enabling dynamic adaptation to local prediction…”
Soft Actor–Critic Learning of Finite State Machine Transition Policies for Lunar Rovers
Rouco-Crenshaw, R., Campbell, A. C., Money, G., & Larson, J. D. (2026) · AIAA SciTech Forum · doi:10.2514/6.2026-1039
CPT: Competence-Progressive Training Strategy for Few-Shot Node Classification
Yan, Q., Zhang, Y., Zhang, J., Duan, J., & Yin, J. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-5640-3_2
Sandoval, A., & Zhang, W. (2026) · AIAA SciTech Forum · doi:10.2514/6.2026-1288
Xu, M., Yang, T., Zhou, T., Chen, Q., Wang, S., & Zhang, Y. (2025) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm66473.2025.11356968
“Hard Negative Mining [5] targets challenging false positives, Focal Loss [6] down-weights easy examples, and Curriculum Learning [9] progressively increases difficulty.”
TACL: Threshold-Adaptive Curriculum Learning Strategy for Enhancing Medical Text Understanding
Ren, M., Yan, Y., Chen, H., Hu, D., Xu, J. T., & Zeng, X. (2025) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm66473.2025.11356814 · Cited by 2
“Recently, curriculum learning [8], a training scheme that aims to let the model learn from easier to more complex samples or tasks, can benefit machine learning algorithms.”
Progressive Learning for Semi-Supervised Ultrasound Image Segmentation
He, G., Xiang, X., Gong, Y., Liu, T., Mei, W., & Tan, S. (2025) · IEEE International Conference on Bioinformatics and Biomedicine (BIBM) · doi:10.1109/bibm66473.2025.11356878
“We address this via curriculum learning [14], re-placing random CutMix with bidirectional region replacement and dynamic reweighting.”
Granular Ball and Curriculum Learning-Based Fuzzy Support Vector Machine
Wu, Z., Pan, X., & Zhou, L. (2025) · IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom) · doi:10.1109/trustcom66490.2025.00174
“Curriculum learning [15]–[17] has opened another path for robustness research in machine learning.”
FakeZero: Real-Time, Privacy-Preserving Misinformation Detection for Facebook and X
Essahli, S., Sarsar, O., Fouad, I., Bentajer, A., & Motii, A. (2025) · IEEE International Conference on Trust, Security and Privacy in Computing and Communications (TrustCom) · doi:10.1109/trustcom66490.2025.00235 · Cited by 2
Q-Forge: An Efficient Two-Stage Framework for Noise-Aware Quantum Model Training
Nguyen, N.-T., & Thai, H.-T. (2025) · RIVF International Conference on Computing and Communication Technologies (RIVF) · doi:10.1109/rivf68649.2025.11365059
“To address these limitations, we propose Q-Forge , a two-stage training strategy inspired by curriculum learning [21] .”
Zhu, S., Niu, K., Li, J., et al. (2026) · Journal of King Saud University - Computer and Information Sciences · doi:10.1007/s44443-026-00528-6
Checkmating One, by Using Many: Combining Mixture of Experts With MCTS to Improve in Chess
Helfenstein, F., Czech, J., Blüml, J., Max, E., & Kersting, K. (2026) · IEEE Transactions on Games · doi:10.1109/tg.2026.3659995 · Cited by 9
“Our staged learning strategy is inspired by the principles from curriculum learning [3] and task decomposition [2], where complex behaviors are learned through progressive specialization and structured modularity.”
Parvathy, S., & Sivasankar, S. (2026) · Journal of Building Pathology and Rehabilitation · doi:10.1007/s41024-026-00787-w
Text-Driven Hybrid Curriculum Learning for Multimodal Sentiment Analysis
Xu, F. X. F., Li, P., Li, D., Huang, L., & Zhu, W. (2026) · Lecture notes in computer science · doi:10.1007/978-981-95-6960-1_19
Gao, X., Hui, T., Yin, X., & Liew, A. W. (2026) · Sensors · doi:10.3390/s26031068
“This is somewhat similar to the ideas of progressive learning [39] and curriculum learning [40].”
Of All StrIPEs: Investigating Structure-Informed Positional Encoding for Efficient Music Generation
Agarwal, M., Wang, C., & Gaël Richard, A. (2026) · IEEE Transactions on Audio Speech and Language Processing · doi:10.1109/taslpro.2026.3662483 · Cited by 1
“Restrictions apply. size of 8, gradient clipping and curriculum learning [38].”
JEONG, Y., Oh, N., Song, B., & Shin, J. (2026) · Transactions of Korean Society of Automotive Engineers · doi:10.7467/ksae.2026.34.2.267
Progressive Text-to-3D Generation for Automatic 3D Prototyping
Han, Y., Zheng, Z., Xu, X., & Chua, T. (2026) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3797262 · Cited by 12
“To update the parameters, we employ a frozen Stable Diffusion model to estimate the injected noise on the rendered image ( e.g. , tiger) and then backpropagate the gradient. similar to curriculum learning [2], where the model starts with simpler tasks and gradually advances to more complex ones.”
ReasonLite: A Unified Framework for Distilling Reasoning into Small Language Models
Thankachan, K. (2025) · International Conference on Ubiquitous Computing and Intelligent Information Systems (ICUIS) · doi:10.1109/icuis67429.2025.11380787
“…reliable research practice: a single configuration schema, global seeding, and run manifests (code/config/environment hashes) for reproducibility; curriculum scheduling to structure learning trajectories [10]; and trace-budget controllers that constrain token usage during training and evaluation.”
InfluentialYe, B., Xia, J., Wan, X., Wu, M., & Chiba, S. (2026) · Methods in Ecology and Evolution · doi:10.1111/2041-210x.70264
“Moreover, curriculum learning could pro - vide a suitable training strategy for multi-barcode data with differ - ent sample coverages, as this approach trains a machine learning model from easier data to harder data (Bengio et al., 2009; Wang et al., 2022 ).”
Metadata
- Authors
- Bengio, Y., Louradour, J., Collobert, R., & Weston, J.
- Affiliations
- Université de Montréal (CA)
- NEC Laboratories America, Princeton, NJ
- ORCID
- Bengio, Yoshua
- Venue
- Proceedings of the 26th Annual International Conference on Machine Learning
- Publisher
- ACM
- Published
- 2009-06-14
- Pages
- 41-48
- OpenAlex
- W2296073425
- Semantic Scholar
- 8de174ab5419b9d3127695405efd079808e956e8
- Cited by
- 5,366
- References
- 40
- Citations by year
- 2026: 526 · 2025: 636 · 2024: 551 · 2023: 591 · 2022: 536 · 2021: 711 · 2020: 565 · 2019: 442 · 2018: 293 · 2017: 198 · 2016: 103 · 2015: 63 · 2014: 41 · 2013: 40 · 2012: 33
- FWCI
- 16.17
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- International Conference on Machine Learning (publishedVersion)
- Queen Mary Research Online (Queen Mary University of London) (submittedVersion)
- Funders
- Canadian Institute for Advanced Research
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Stochastic Gradient Optimization Techniques
- Physical Sciences › Computer Science › Artificial Intelligence › Reinforcement Learning in Robotics
- Physical Sciences › Computer Science › Artificial Intelligence › Neural Networks and Applications
- Keywords
- curriculum learning, training order, non-convex optimization, deep neural network, stochastic neural networks, deterministic neural networks, generalization, convergence rate, numerical continuation, global optimization
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.