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 5,101–5,200 of 5,240
Ray Augmented Supervision for 3D Object Detection
Duong, H.-H., Voicila, A., & Allibert, G. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-31673-8_23
Weighted Meta-Learning for Stealthy Poisoning Attacks on Graph Link PredictionPreprint
Li, M., Zhuang, D., Chen, K., et al. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7228090
Metamodel-Driven Modeling of UAF-Based Cooperative Drone Combat Systems with AutoCL-MAPPO
Feng, Y., Li, Y., zhang, P., Chen, J., Zhao, G., Liu, Y., Li, H., & Huang, Y. (2026) · Systems · doi:10.3390/systems14080927
A Trainable Connected Filter Preprocessing Layer Based on Component Trees
Alves, W. A. L., Santos, L. D. P. O., Hashimoto, R. F., Passat, N., Souza, A. H. R., Silva, D. J., & Kenmochi⋆, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-31654-7_36
CG-CLIP: Seeing Beyond Objects to Improve OOD Corruption Detection
Jain, D., Ali, A., & Majumder, A. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-31335-5_39
Multimodal History-Window Gated-Attention Soft Actor-Critic for Urban Low-Altitude UAV Navigation
You, X., & Yi, W. (2026) · Drones · doi:10.3390/drones10080605 · Cited by 1
Divide and Conquer: Region Self-Paced Physics-Informed Neural Networks
Meng, R., Wu, W., Ren, Z., et al. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3818114
Distribution-Value Coevolution for Adaptive RLHF Data Scheduling
Yang, Z., Yang, Y., Zhou, C., Guan, X., Dong, B., Zhang, M., Ding, K., Zhu, H., Chen, H., & Zhang, Q. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3817990
Charting the small-molecule universe from mass spectra with neuro-symbolic AIPreprint
Acikalin, U. U., Feng, D., Ferber, A., Gouveia, G. J., Schwertfeger, T. J., Chen, D., Qu, D., Fontaine, M. A., Wang, Y., Bernstein, R. A., et al. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.08.05.743095 · Cited by 1
“To stabilize optimization over the highly combinatorial fragmentation space, DeepMS 2 Reasoner is trained using a curriculum strategy 39 that gradually introduces higher-mass molecules.”
Enhancing Protein Representation Learning via Manifold Restore Mixing
Dang, Y., Zhao, C., Ma, L., Guo, G., Wang, X., & Sun, Z. (2026) · ACM SIGKDD Conference on Knowledge Discovery and Data Mining (KDD) · doi:10.1145/3770855.3818845
“Existing research has demonstrated that training samples arranged from easy to difficult can effectively enhance model performance and robustness [3, 63].”
An Empirical Study on Curriculum Learning for Reinforcement Learning-based Biomechanical Arm Control
Ouhssain, A., Ibrahim, M.-H., Harispe, S., et al. (2026) · International Conference on Control, Decision and Information Technologies (CoDIT) · doi:10.1109/codit70676.2026.11630991
“CL refers to a training strategy in which learning samples or tasks are presented in a structured order rather than being sampled randomly [8].”
Learning Faster without Deeper Networks: A*-Inspired Batch Selection for Efficient CNN Training
Shehu, A., Stastoli, E., & Çela, A. (2026) · International Conference on Control, Decision and Information Technologies (CoDIT) · doi:10.1109/codit70676.2026.11630921
“A significant body of work has focused on curriculum learning and self-paced learning strategies, where training samples are presented in a meaningful order to improve convergence and stability, as introduced by Bengio et al. [5] and later extended by Kumar et al. [6].”
GEM-Distill: Graph Ensemble Multi-Level Distillation for MLP-Based Graph Classification
Qin, M., Guan, D., Yuan, W., Xu, S., & Koç, Ç. K. (2026) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2026.3722236
“Inspired by curriculum learning [61] and AdaBoost-style reweighting [60], [62], these experts are organized into a coarse-to-fine sequential framework within each training epoch.”
Latent Aging Dynamics Network for Multi-State Time Series Prediction
Aksu, O., Schmid, M., Kucharczyk, A., Janeva, J., Bierwirth, F., Radosavac, M., & Herzog, H. (2026) · IEEE International Conference on Intelligent Reality (ICIR) · doi:10.1109/icir70631.2026.11628124
“This progressive strategy draws on the principle that structured, stage-wise training yields better dis-entanglement than simultaneous optimization [32], [34], [37].”
Learning Hierarchies in Robotic Manipulation: Asymmetries in Sequential Pushing and Grasping
Mbua, P. N., Chen, J., Wang, X., & Gao, Z. (2026) · IEEE China Conference on System Simulation Technology and its Applications (CCSSTA) · doi:10.1109/ccssta69471.2026.11635785
Iturrarán‐Viveros, U. (2026) · Ingeniería Investigación y Tecnología · doi:10.22201/fi.25940732e.2026.27.3.023
“Instead of directly learning the target PDE with its full complexity, training starts with easier versions (e.g., smaller coefficients, shorter time domains) and gradually increases to the target values, following the curriculum learning principle (Bengio et al. , 2009)”
Kwon, J. (2026) · IEEE Access · doi:10.1109/access.2026.3722619
“Curriculum learning [24] orders or weights training samples so that informative or progressively harder examples drive optimization, and importance sampling [25] formalizes the same intuition by sampling examples in proportion to their contribution to the loss, improving convergence and final…”
FedCurrMM: A Federated Map Matching Framework with Curriculum-Aware Client Selection
Chen, M., Yuan, H., Wang, H., Jiang, N., Zheng, Z., Zhou, A., & Wang, S. (2026) · IEEE International Conference on Data Engineering (ICDE) · doi:10.1109/icde65706.2026.00116
“Unlike existing approaches that rely on short-term heuristics or local feedback, our method is inspired by curriculum learning [22] and views client selection as a dynamic curriculum scheduling process.”
Information-Theoretic Limits of Scenario-Based Validation for Agentic Artificial Intelligence
S, S. (2026) · i-manager s Journal on Computer Science · doi:10.26634/jcom.14.1.11256
An SPCL-Stacking ensemble learning framework for landslide susceptibility assessment
Zhou, H., Li, L., Wang, X., Huang, B., Chen, K., & Zhang, S. (2026) · Geoenvironmental Disasters · doi:10.1186/s40677-026-00406-3
“Their common characteristic is that simple samples are introduced first during training, followed by more complex samples, thereby improving the learning performance in nonconvex optimization processes (Bengio et al. 2009, 2013; Wang et al. 2022).”
STQFormer: Spatio-Temporal Quaternion Transformer for Video Frame Denoising
Jin, A., Zhang, S., & Wang, S. (2026) · Proceedings - International Conference on Image Processing · doi:10.1109/icip61757.2026.11630495
Zhang, W., Deng, J., Wu, Z.-Y., et al. (2026) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2026.3723486
“The dynamic scheduling of the memory length M follows the curriculum learning paradigm [45] under a coarse-to-fine principle: M expands the memory horizon in the early training stage to explore global long-range patterns, and contracts later to focus on fine-grained local features.”
Ahmad, I., Khubaib, M., & Altowaijri, S. M. (2026) · Diagnostics · doi:10.3390/diagnostics16162571
“Data-level imbalance mitigation has been strengthened by generative models based on Generative Adversarial Networks (GANs), which have been used to create realistic synthetic medical images of minority classes, such as DCGAN, Conditional GAN (cGAN), BSS-GAN and DB-cGAN [24,32,33].”
In-context curriculum learning for enhanced semantic understanding in scientific text mining
Liu, Y., Liu, J., Shi, X., Huang, Y., Lu, W. P., & Cheng, Q. (2026) · Scientometrics · doi:10.1007/s11192-026-05767-y
HGNNMap: Heterogeneous Graph Neural Network-Based Mapping for Spatial Accelerators
Tang, S., Fan, Z., Liu, T., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-35248-4_19
Entropy-Based Gradient Weighting and Batch-Size Adaptation for Virtual Data-Parallel Training
Sun, T.-J., & Huh, E. (2026) · Communications in computer and information science · doi:10.1007/978-981-92-2597-2_24
Improved maskable PPO for budget-constrained data-collection path planning
Miao, J., Liu, G., Wu, S., Zhu, L., Yu, W., & Zhou, Q. (2026) · Cluster Computing · doi:10.1007/s10586-026-06488-w
“The curriculum begins with smaller environments and gradually increases the grid size and target number [ 39].”
Evolution of humanoid locomotion control
Gu, Y., Shi, G., Shi, F., Chang, I.-C., Wang, Y., Cheng, Q., Olkin, Z., López-Sánchez, I., Feng, Y., Zhang, J., et al. (2026) · Science Robotics · doi:10.1126/scirobotics.aed3973 · Cited by 8
Talib, M. W. A., Nasir, Q., Ali, S. K., et al. (2026) · Array · doi:10.1016/j.array.2026.101146
Randomize and Identify for Stable Flight
Boudlal, A., Khafaji, A., & Elabbadi, J. (2026) · Lecture notes in networks and systems · doi:10.1007/978-3-032-28097-8_17
Mobility tokens with routine-aware learning for LLM-based human mobility prediction
Wang, C., Ge, Q., Hu, N., et al. (2026) · GeoInformatica · doi:10.1007/s10707-026-00589-9
“The supervised fine-tuning loss is then scheduled from easier, routine-consistent slots toward harder, irregular slots, following the principle of curriculum learning [21].”
InfluentialLi, Z., Zhang, J., Zhou, X., Zhang, M., Chen, Z., Gao, G., & Su, X. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-36033-5_8
Zhang, M., Zhou, X., Li, Z., et al. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-36033-5_35
Bir, U. B., Christlein, V., Maier, A., & Zinnen, M. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-36039-7_30
“We investigate whether curriculum-based multi-stage training [5] improves convergence and final performance compared to single-stage fine-tuning.”
Chang, H., Wang, X., Yu, H., Duan, Y., Wang, K., Gu, Y., Zhang, L., Tian, M., & Pei, Y. (2026) · Applied Sciences · doi:10.3390/app16178395
“Curriculum learning and hard-example mining change training exposure according to difficulty [15,16].”
Curriculum Self-Supervised Learning for BD-RIS Optimization in Multi-User MIMO Systems
Pérez-Adán, D., Pereira-Ruisánchez, D., Fresnedo, Ó., et al. (2026) · International Symposium on Communication Systems, Networks and Digital Signal Processing (CSNDSP) · doi:10.1109/csndsp68462.2026.11654391
Collaborative Task-Oriented Communication-Control Co-Design for Wireless Networked Control Systems
Zhang, L., Zheng, M., Liang, W. G., et al. (2026) · IEEE Transactions on Cognitive Communications and Networking · doi:10.1109/tccn.2026.3726548
“To enhance training stability of the control model, we employ a two-stage curriculum learning strategy [35].”
Staged Fine-Tuning of Large Language Models for Multi-Level Space Station Operation Mission Planning
Xu, L., Ding, R., Huang, X., Zhou, Y., He, Y., & Xu, Y. (2026) · Aerospace · doi:10.3390/aerospace13090757
“…few high-IL samples to form schemata and then progressively expands to simpler, higher-volume samples—a strategy that extends Curriculum Learning [27] by replacing monotonically increasing difficulty with a three-dimensional cognitive load framework that explicitly maps each stage to a domain…”
Mai, S., Zhang, Z., Zhu, J., Mai, W., & Deng, G. (2026) · Journal of King Saud University - Computer and Information Sciences · doi:10.1007/s44443-026-01075-w · Cited by 1
“…curriculum learning pipeline encompassing masked language modeling on a 4.2-billion-token climate corpus, contrastive pre-training with a purpose-built 18,463-entry bilingual climate disclosure lexicon, and task-specific fine-tuning on the C3D dataset (Chen et al. 2020; Bengio et al. 2009).”
InfluentialCurriculum-Based Deep Reinforcement Learning for Robust Biped Push Recovery
Ercolanese, L., Aliotta, R., Arpenti, P., Ruggiero, F., & Lippiello, V. (2026) · IEEE/ASME International Conference on Advanced Intelligent Mechatronics · doi:10.1109/aim65483.2026.11658159
“Curriculum learning [21] addresses this issue by progressively increasing task difficulty according to training progress, thereby improving learning efficiency and accelerating convergence.”
Optimizing Training Data Diversity with DataSpinner: An LLM-driven Curator-Agent Approach
Horti, D., Sopkovič, K., Hládek, D., & Karoly, A. I. (2026) · IEEE International Conference on Intelligent Engineering Systems (INES) · doi:10.1109/ines69513.2026.11661205
“In Curriculum Learning, examples are presented in an order (often easy-to-hard) that accelerates optimization and improves final performance [20].”
Pitre, T., Marques, L., Weatherald, J., Mak, S. S., Thavendiranathan, P., & Granton, J. T. (2026) · medRxiv · doi:10.64898/2026.08.19.26360852
Zhu, W., Tu, J., Huang, H., He, Z., Xie, Z., Li, L., & Yao, J. (2026) · Electronics · doi:10.3390/electronics15173844
Ünaldı, A. O., Yılmaz, M., & Clarke, P. M. (2026) · Communications in computer and information science · doi:10.1007/978-3-032-36816-4_17
REMIX: Adaptive Open-Domain Signal Mixing for Efficient Sensing Data Augmentation
Kara, D., Kimura, T., Hu, C., Li, J., Chen, Y., Hu, Y., Zhao, H., Liu, S., & Abdelzaher, T. (2026) · International Conference on Computer Communications and Networks (ICCCN) · doi:10.1109/icccn69946.2026.11662941
Mental bootstrapping enables human-level concept learning in self-supervised deep models
Yang, L., Lyu, M., Wang, Y., Xie, X., Zhen, Z., Lai, J., & Zhang, R. (2026) · Science Advances · doi:10.1126/sciadv.aea7202
“For example, in natural language processing, models may benefit from first acquiring partial syntactic or semantic regularities before being exposed to full linguistic complexity, mirroring human language acquisition ( 2 , 47 ).”
Alkhodre, A. B., Al-Amine, M. A., & Alsaawy, Y. (2026) · Drones · doi:10.3390/drones10090660
“Curriculum learning [19] structures the training process by gradually increasing task difficulty.”
Jiang, L., Chen, J., Lan, K., et al. (2026) · Biomimetics · doi:10.3390/biomimetics11090610
“Building on these biomimetic principles, this paper presents a robotic piano-playing framework based on a two-stage reinforcement learning curriculum [25–27].”
Active Noise Control Using Reinforcement Learning with TD3 for Suppressing Narrowband Noise
Kim, J., & Lee, Y. (2026) · International Journal of Automotive Technology · doi:10.1007/s12239-026-00543-6
FlightLLM: few-shot interval forecasting for airline passenger and baggage demand
Gao, Y., Li, C., Zhong, Z., & Ma, W. (2026) · Transportation Research Part C Emerging Technologies · doi:10.1016/j.trc.2026.105941
Chen, Y., Liu, Y., Lei, N., et al. (2026) · Green Energy and Intelligent Transportation · doi:10.1016/j.geits.2026.100472
Augmenting End-to-End Congestion Control with Centralized Reinforcement Learning
König, M., & Zitterbart, M. (2026) · IEEE Network Operations and Management Symposium (NOMS) · doi:10.1109/noms69089.2026.11668338
“To stabilize training, we employ curriculum learning [13] by progressively increasing scenario complexity.”
Melo, D. A. D. J., Medeiros, P. A. G., Schmitt, L. E. M., et al. (2025) · doi:10.5753/sbcm.2025.13240
Tu, C.-R., Hsueh, H.-W., Chang, W.-C., Lin, Y.-T., Cheng, Y.-H., Chu, W.-H., Tseng, S.-W., & Hsu, S.-H. (2026) · ACS Measurement Science Au · doi:10.1021/acsmeasuresciau.6c00264
Zhuo, Y., Ren, Y., Zhang, Y., & Wang, L. (2026) · Applied Sciences · doi:10.3390/app16178778
“Curriculum learning progressively increases the difficulty of source–destination pairs [34]: Stage 1 (training progress 0–10”
Palmer, P., Krüger, M., & Bertram, T. (2026) · Proceedings of the IEEE National Radar Conference · doi:10.1109/radarconf2663773.2026.11675387
“The MSTM is inspired by curriculum learning [20].”
An efficient self-paced multi-view method for partial label learning
Liu, B., Chen, D., Xiao, Y., et al. (2026) · Knowledge-Based Systems · doi:10.1016/j.knosys.2026.116969
Jiang, J., Yu, D., Li, C., Wang, D., Nishizaki, H., & Leow, C. S. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37670-1_11
Gao, S., Zhou, F., & Shao, S. (2026) · Electronics · doi:10.3390/electronics15174046
DualGLEAN: Dual Allocation for VLM-Guided Generalized Category Discovery in Remote Sensing Images
Li, H., Xie, Y., Zhang, J., et al. (2026) · Remote Sensing · doi:10.3390/rs18173054
Cooperative Multi-Agent Reinforcement Learning for Idle-Mode Cell (Re)Selection
Illian, M., Khalili, R., Rocha, A. A. D. A., et al. (2026) · Conference on Local Computer Networks · doi:10.1109/lcn67947.2026.11660770
Intentional Degradation for Stabilizing Residual Learning in Coarse-to-Fine Medical Image Refinement
Kim, M., Lee, S., Kim, H., Park, S.-C., Park, J., Yoon, J., Yong, S.-C., Hur, J., & Kym, D. (2026) · Journal of Imaging · doi:10.3390/jimaging12090418
“Intentional Degradation addresses this by explicitly compressing the target distribution toward the coarse prediction space—a strategy conceptually related to, yet directionally inverse to, curriculum learning [10,11], which adjusts input complexity rather than target distribution.”
SCALER: LLM-Based Cross-Modal Alignment for Microservice Root Cause Analysis
Yang, S., Wang, G., Ming, Z., He, D., Cui, L., Yu, X., & Zhou, X. (2026) · IEEE International Conference on Web Services (ICWS) · doi:10.1109/icws72778.2026.00101
“Curriculum learning has shown effectiveness across various domains [43] [44] [45].”
ACR-Nav: Localization-Free Corridor Navigation via Action-Conditioned Scalar-Range Evolution
Shen, Q., Wang, Z., Feng, Y., & Xu, K. (2026) · Machines · doi:10.3390/machines14091026
Knowledge-Guided Relational Reasoning for Robust Network Topology Optimization
Chen, W., Zhang, S., Lu, B., et al. (2026) · IEEE International Conference on Smart Internet of Things (SmartIoT) · doi:10.1109/smartiot70864.2026.00012
“To reduce the reward variance in the early training stage, this paper adopts a two-stage training strategy inspired by curriculum learning [22].”
Lee, J., Bak, G.-M., & Bae, Y. (2026) · IEEE Access · doi:10.1109/access.2026.3731873
“The MAB + PPO curriculum handoff operates as follows [20].”
From Retained Change to Learning,A reverse reconstruction of event-driven changePreprint
Hintsanen, A. D. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7434345
Sangwan, M., Bajpai, S., Gupta, A., et al. (2026) · IEEE Access · doi:10.1109/access.2026.3732194
“Bengio et al. [20] show that, when training models to progressively hard tasks, it yields better performance.”
MPO: Single-Stream Policy Optimization for Efficient Text-to-Image Alignment
Gao, L., Hu, J., Wei, C., Zhong, Y., Zhao, Y., Gao, Z., & Wei, X. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37359-5_4
Wavelet-Based Intra-Video Counterfactual Reasoning for Video Question Grounding
Yuan, J., & Wei, W. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37098-3_35
Zai, W., Li, K., Yan, L., & Dai, Y. (2026) · Optical Review · doi:10.1007/s10043-026-01081-3
RecGPT: A User Intent-Centric Next-Generation LLM-Powered Recommender System in Industrial Practice
Tang, J., Chen, W., Chen, D., Chao, Y., Guo, G., Yang, W., Yu, J., Gao, Z., Zhang, M., Dai, S., et al. (2026) · ACM Transactions on Information Systems · doi:10.1145/3846382
“To ensure stable capability enhancement, inspired by curriculum learning principles [6, 34, 48, 62], we organize subtasks through topological sorting based on difficulty levels and dependency relationships, progressively guiding the model to master complex tasks.”
DFRL-Mol: a dual-stage framework of reinforcement learning for multi-scenario molecule optimization
Wu, Z., Zhao, P., Li, J., Wang, Q., Zhang, Q., Yu, H., Yu, Z., & Shi, J. (2026) · Briefings in Bioinformatics · doi:10.1093/bib/bbag493
“…obj , all other components function as binary indicators, yielding 1 if the requirement is met and 0 otherwise. reward obj is defined as follows: To address the issue of reward sparsity caused by the lack of high-quality data, DFRL-Mol introduces the Curriculum Learning (CL) method [ 24, 25].”
3D UAV Path Planning and Residual PPO Dynamic Obstacle Avoidance for Substation Inspection
Lv, M., Yang, Z. X., Liao, L., et al. (2026) · International Conference on Computer Information Science and Application Technology (CISAT) · doi:10.1109/cisat71030.2026.11682299
“Curriculum learning [9] gradually increases dynamic obstacles and unknown blockages.”
RF-FilterLLM: A Fine-Tuned LLM Agent for Automated RF Filter Design
Sun, W., Li, Y., Wang, Z., Sui, X., Luo, X., Mei, Y., Li, Q., Wang, R., Shi, H., & Pang, J. (2026) · International Conference on Microwave and Millimeter Wave Technology (ICMMT) · doi:10.1109/icmmt69626.2026.11678389
“Curriculum-ordered training [6] arranges samples by a composite difficulty score (order complexity, parameter extremity, and band type) with intra-bucket shuffling to avoid degenerate orderings.”
DSHM: Dynamic Semantic Hard Mining for Text-Based Person Re-Identification
Huang, B., Zhang, H., & Ma, Z. (2026) · International Conference on Computer Information Science and Application Technology (CISAT) · doi:10.1109/cisat71030.2026.11682126
“This strategy follows the curriculum learning paradigm and gradually increases the training difficulty from easy samples to hard samples[5].”
SDUM: A Scalable Deep Unrolled Model for Universal Cardiac MRI Reconstruction
Wang, P., Guo, P., Chai, K., Zhou, J., Xu, D., & Jiang, S. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37435-6_21
A Model-Driven Approach for Developing Families of Reinforcement Learning Environments
Liu, X., & Dávid, I. (2026) · ACM/IEEE International Conference on Model Driven Engineering Languages and Systems (MODELS) · doi:10.1145/3822455.3830332 · Cited by 1
Chen, Y., Liu, Y., Lei, N., et al. (2026) · Green Energy and Intelligent Transportation · doi:10.1016/j.geits.2026.100470
PINCH: Predictive Importance-Sampling-Informed Cache for I/O-Bound DNN Training
Liu, Y., & Yin, S. (2026) · Proceedings of the International Conference on Parallel Processing · doi:10.1145/3832810.3832851
Interpretable and Robust Car-Following Planning via Causality-Driven Deep Reinforcement Learning
Han, Z., Li, Y., Chen, K., et al. (2026) · Lecture notes in electrical engineering · doi:10.1007/978-981-92-3570-4_65
Hi-DiT: Hybrid Latent-Pixel Diffusion Transformer for Image Generation
Shen, Y., Li, Y., Pan, Y., Zhang, Y., & Yao, T. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37255-0_11
Adaptive Emission Trading System Policy Design via Bi-level Multi-agent Reinforcement Learning
Long, X., Zeng, X., & Meng, F. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-38404-1_11
Alqadhi, A. M. N. S., Muhammed, M. E., Bajhaw, M. A. M. S., & Uçar, A. (2026) · Applied Sciences · doi:10.3390/app16189083
Song, P., Wu, J., & Xiao, Y. (2026) · Communications in Mathematical Research · doi:10.4208/cmr.2026-0066
Hong, J., Park, J., & Kwon, J. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37324-3_10
Wang, L., Huang, Q., Wang, K., LOU, J., & Yuan, C. (2026) · International Journal of Human-Computer Interaction · doi:10.1080/10447318.2026.2728615
A deep reinforcement learning framework for coastal multi-hazard emergency response
Sanò, M., Ferrario, D. M., Torresan, S., & Critto, A. (2026) · International Journal of Disaster Risk Reduction · doi:10.1016/j.ijdrr.2026.106439
Terrain-Aware Reinforcement Learning for Unified Lower-Limb Flat-Ground Walking and Stair Ascent
Wu, Z., Wang, W., Luo, Y., et al. (2026) · doi:10.1109/mesa70578.2026.11684238
“To facilitate the learning process, a staged curriculum first establishes stable flat-ground locomotion and then progressively introduces stair episodes with geometry-compatible commanded velocities [4].”
Enhancing Small Model Information Extraction with Few-Shot Data: A Dynamic Teaching Approach
Du, W., Zhang, Y., Zhang, D., et al. (2026) · International Conference on Big Data and Information Analytics (BigDIA) · doi:10.1109/bigdia71496.2026.11684643
“…provide guidance and which model should receive improvement at different learning stages [4]; (3) Insufficient consideration of sample difficulty: conventional teaching strategies often rely on predefined difficulty levels and may not fully consider the specific weaknesses of individual models [5], [6].”
Zhou, Z., Xin, H., Tang, P., et al. (2026) · International Conference on Big Data and Information Analytics (BigDIA) · doi:10.1109/bigdia71496.2026.11684697
“To enhance robustness against background noise, we pro-pose a Curriculum Learning-based noise injection strategy [18].”
Nie, N., Zhu, X., Zang, Y., & Tang, H. (2026) · International Conference on Big Data and Information Analytics (BigDIA) · doi:10.1109/bigdia71496.2026.11684618
“We train with a three-stage curriculum [21]: basic tracking for 3000 iterations, obstacle tracking with 15 obstacles for 3000 iterations, and dense-obstacle compound-disturbance tracking with 25 obstacles, 2.0 m/s crosswind, and asymmetric torque bias for 4000 iterations.”
InfluentialJo, Y. R., & Lee, J. Y. (2026) · IFIP advances in information and communication technology · doi:10.1007/978-3-032-38606-9_7
Curriculum-driven reinforcement learning for autonomous pre-capture of dynamic space debris
Beigomi, B., Makhdoomi, M. R., Olivares-Méndez, M. A., Martínez, C., & Zhu, Z. H. (2026) · Robotics and Autonomous Systems · doi:10.1016/j.robot.2026.105752
Tag Entropy: A Structure-Agnostic Curriculum Learning Framework for Graph Neural Networks
Madhusanka, D., Mohottala, S., Gawesha, A., Kommalage, S., Weerasinghe, M. M. A., Kasthurirathna, D., & Abhayaratne, C. (2026) · Moratuwa Engineering Research Conference (MERCon) · doi:10.1109/mercon71835.2026.11691316
“…training paradigm that re-places the random sample ordering of standard neural network training with a structured progression from easy to hard examples, inspired by the observation that humans and animals acquire knowledge more effectively when concepts are introduced from simple to complex [10].”
TiCRL: Textual Image Classification with Reinforcement Learning-Based Curriculum Learning
Kim, G., & Kang, Y. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-37321-2_23
Multi-Step Obstruction Reasoning for Target-Oriented Grasp Sequence Generation in Cluttered Scenes
Yang, H., Zhu, S., Zheng, Y., & Song, W. (2026) · Robotics · doi:10.3390/robotics15100181
MPDRL: difficulty-aware representation learning with medical priors for vision–language pre-training
Liu, J., Tan, Z., Chen, W., & Liu, X. (2026) · Frontiers in Artificial Intelligence · doi:10.3389/frai.2026.1906686
Barhoumi, D., Hensel, S., & Marinov, M. B. (2026) · Applied Sciences · doi:10.3390/app16189359
Shin, J. W., Morris, C. A., Chee, F. S., et al. (2026) · International Conference on Pattern Recognition and Machine Learning (PRML) · doi:10.1109/prml71097.2026.11687118
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.