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 4,101–4,200 of 5,240
Mapless Navigation: Learning UAVs Motion forExploration of Unknown EnvironmentsPreprint
Jung, S., & Shim, D. H. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.01747 · Cited by 1
“Thus, to solve this problem, the method of curriculum learning [25] is introduced.”
MethodSimple Baseline for Single Human Motion ForecastingPreprint
Wang, C., Wang, Y., Huang, Z., et al. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2110.07495 · Cited by 1
BAMLD: Bayesian Active Meta-Learning by DisagreementPreprint
Nikoloska, I., & Simeone, O. (2021) · arXiv (Cornell University) · Cited by 1
SDCUP: Schema Dependency-Enhanced Curriculum Pre-Training for Table Semantic ParsingPreprint
Qin, B., Wang, L., Hui, B., Geng, R., Cao, Z., Yang, M., Sun, J. Q., & Li, Y. (2021) · arXiv (Cornell University) · Cited by 1
Meta Arcade: A Configurable Environment Suite for Meta-LearningPreprint
Staley, E. W., Ashcraft, C., Stoler, B., Markowitz, J., Vallabha, G. K., Ratto, C. R., & Katyal, K. D. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2112.00583 · Cited by 6
“If a target environment is difficult to learn but can be modified to create more easily-learned variants, it can be useful to supply a learning curriculum [19, 20, 21, 22].”
BackgroundThe role of motion-and-visual perception in robot place learning and navigation
LEON, M. A. C. (2022) · Queensland University of Technology · doi:10.5204/thesis.eprints.229769 · Cited by 2
CityLearn: Diverse Real-World Environments for Sample-Efficient Navigation Policy LearningPreprint
Chancán, M. (2020) · doi:10.36227/techrxiv.12063582.v1 · Cited by 1
Abstractive Turkish Text Summarization and Cross-Lingual Summarization Using Transformer
Taspinar, E. K., Yetis, Y. B., & Cihan, O. (2022) · Advances in computational intelligence and robotics book series · doi:10.4018/978-1-6684-6001-6.ch011
Self-Paced Contrastive Learning for Knowledge TracingPreprint
Dai, H., Yun, Y., Zhang, Y., et al. (2023) · SSRN Electronic Journal · doi:10.2139/ssrn.4627338 · Cited by 1
Li, J., Xiao, L., Luo, J., et al. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.12.16.570150 · Cited by 1
“To end these issues, we introduced the domain constraint [19] and curriculum learning [20] in the feedback loop ( Fig.”
MethodHocker, D. L., Constantinople, C. M., & Savin, C. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.01.12.575461 · Cited by 1
“Shaping is the process by which an animal is gradually exposed to (and thus learns) the core components of a behavioral task, such as whether or not to interact with effectors like levers or ports, or associate sensory cues with rewards or abstract rules, based on reinforcement.”
BackgroundPhysics-Informed Reinforcement Learning Optimization of Pwr Core Loading PatternPreprint
Seurin, P., & Shirvan, K. (2024) · SSRN Electronic Journal · doi:10.2139/ssrn.4772551 · Cited by 1
Kurmi, Y., Viswanathan, M., & Zu, Z. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.06.07.597818 · Cited by 1
Predicting human prediction error empowers reward learning task designPreprint
Shin, J., Lee, J. H., & Lee, S. W. (2024) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2024.12.08.627441 · Cited by 2
“One key area is curriculum learning 49 , where our framework enhances task sequencing by streamlining a progression of learning experiences that promote goal-directed behavior without compromising overall learnability.”
LLMulator: Generalizable Cost Modeling for Dataflow Accelerators with Input-Adaptive Control Flow
Chang, K., Zhu, W., Liang, S., Li, H., & Wang, Y. (2025) · IEEE/ACM International Symposium on Microarchitecture (MICRO) · doi:10.1145/3725843.3756027
Graph-based deep reinforcement learning for haplotype assembly with Ralphi
Battistella, E., Maheshwari, A., Ekim, B., Berger, B., & Popic, V. (2025) · Genome Research · doi:10.1101/gr.280569.125 · Cited by 2
The Autonomy-Alignment Problem in Open-Ended Learning Robots Formalising the Purpose FrameworkPreprint
Baldassarre, G., Duro, R. J., Cartoni, E., Khamassi, M., Romero, A. M., & Santucci, V. G. (2025) · SSRN Electronic Journal · doi:10.2139/ssrn.5547070 · Cited by 1
Wang, H., Zhang, X., Fang, S., Li, R., Deng, Z., Zhang, Y., Li, Y., & li, S. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.01.09.698573 · Cited by 1
Jiang, G., Li, J., Ma, X., & Luo, Q. (2026) · ACM International Conference on Intelligent User Interfaces (IUI) · doi:10.1145/3742413.3789094
“Generating training cases adapted to previous performance can enhance the effectiveness of training [8, 22].”
Approches par apprentissage pour l’estimation de mouvement multiframe en vidéo
Godet, P. (2021) · theses.fr (ABES) · doi:10.70675/d6ba9ed5z20dbz4620zbe4dz0967f2b8972e · Cited by 2
Zhang, S., Gu, Y., Cater, K. F., & Metatla, O. (2026) · CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791927 · Cited by 2
“This structuring datasets approach, proven in related machine-learning studies, builds robustness and reduces brittleness when faced with out-of-distribution expressions [5, 95], enables models to progressively generalise from stable exemplars to more diverse and atypical executions.”
A Comparative Evaluation of Curriculum Learning with Filtering and BoostingPreprint
Smith, M. R., & Martinez, T. R. (2013) · arXiv (Cornell University) · doi:10.48550/arxiv.1312.4986
“Bengio et al. (2009) formalized these ideas in curriculum learning.”
MethodExploring Applications of Representation Learning in Nepali
Nepal, A., & Yates, A. (2014) · Lecture notes in computer science · doi:10.1007/978-3-642-54906-9_12
Curriculum based discriminative language model training
Dikici, E., & Saraçlar, M. (2013) · Signal Processing and Communications Applications Conference (SIU) · doi:10.1109/siu.2013.6531443
“[9] Yoshua Bengio, Jérôme Louradour, Ronan Collobert, and Jason Weston, “Curriculum learning,” in Proceedings of the 26th Annual International Conference on Machine Learning, New York, NY, USA, 2009, ICML ’09, pp.”
BackgroundNew Paradigms for Active Learning
Ni, A. (2012) · Scholarship@Western (Western University)
Message passing with relaxed moment matchingPreprint
Qi, Y., & Guo, Y. (2012) · arXiv (Cornell University) · doi:10.48550/arxiv.1204.4166
Weakly supervised neural networks for Part-Of-Speech tagging
Chopra, S., & Bangalore, S. (2012) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2012.6288291
“In order to accomodate the fact that it is a weakly supervised training task, we propose a novel method for training the model based on stochastic gradient descent, which exploits a regimented learning strategy called Curriculum Learning [9].”
BackgroundMethodEnhancing agent safety through autonomous environment adaptation
Rosman, B., Hayes, B., & Scassellati, B. (2015) · Joint IEEE International Conference on Development and Learning and Epigenetic Robotics · doi:10.1109/devlrn.2015.7346143
“The benefit of interactive shaping, the practice of providing targeted feedback to manipulate portions of an agent’s policy to facilitate goal achievement, is well established [11], [12], [13].”
BackgroundACM Multimedia (ACMMM) 2014 Report
Kagaya, H., & Yamasaki, T. (2015) · The Journal of The Institute of Image Information and Television Engineers · doi:10.3169/itej.69.141
Neural Associative Memory for Dual-Sequence Modeling
Weissenborn, D. (2016) · Workshop on Representation Learning for NLP (RepL4NLP) · doi:10.18653/v1/w16-1630 · Cited by 1
“We believe that this might be achieved via regularization or by curriculum learning (Bengio et al., 2009).”
BackgroundMachine Teaching for Bayesian Learners.Preprint
Zhu, X. (2013) · arXiv (Cornell University)
Ratinov, L. (2012) · Illinois Digital Environment for Access to Learning and Scholarship (University of Illinois at Urbana-Champaign)
Optimizing machine translation by learning to search
Melamed, D., & Galron, D. (2012) · ProQuest Demo Repository
Improving First Order Temporal Fact Extraction with Unreliable Data
Luo, B., Feng, Y., Wang, Z., & Zhao, D. (2016) · Lecture notes in computer science · doi:10.1007/978-3-319-50496-4_21
“Instead of adding less reliable data directly to the training set, an alternative way to incorporate these data is to use the idea of curriculum learning [3]: start with easier aspect of the task and then increase the difficulty level gradually.”
MethodYang, M., Hsu, Y.-C., Hung, H.-T., Chen, Y., Chen, K., & Chen, B. (2016)
Yang, M., Hsu, Y.-C., Hung, H.-T., Chen, Y., Chen, B., & Chen, K. (2016) · International Conference on Computational Linguistics
Starting small learning strategies for speech recognition
Vydana, H. K., Srivastava, B. M. L., Shrivastava, M., et al. (2016) · IEEE India Conference · doi:10.1109/indicon.2016.7838940
“The basic idea of curriculum is to start small, learn easier examples and gradually increase the difficulty level [2].”
BackgroundInfluentialLearning From Natural Human Interactions For Assistive Robots
Jain, A. (2016) · eCommons (Cornell University) · doi:10.7298/x4mw2f2v
“This simulates curriculum learning [16] and helps in keeping the forecasted motion close to the manifold of human motion.”
BackgroundLearning Slowly To Learn Better: Curriculum Learning for Legal Ontology Population
Cardellino, C., Teruel, M., Alemany, L. A., et al. (2017) · Repositorio Digital de la UNC (National University of Cordoba)
A Practical and Conceptual Framework for Learning in Control
Deisenroth, M. P., & Rasmussen, C. E. (2010) · Spiral (Imperial College London) · doi:10.25561/12654 · Cited by 1
Active Collaborative Ensemble TrackingPreprint
Meshgi, K., Mirzaei, M. S., Oba, S., et al. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1704.08821
“Different studies have tried to provide this small clean subset by different approaches: pruning outliers [12] and hard-to-learn samples [1], learning easy-to-classify examples first (as known as the Curriculum learning) [8], treating samples as noisy observations [14], defining a training value for each sample by tre…”
BackgroundA Conversation with the CHCCS/SCDHM 2016 Achievement Award Winner
van de Panne, M., & Kry, P. G. (2016) · Graphics Interface · doi:10.5555/3076132.3076134
Joint Saliency Estimation and Matching using Image Regions for Geo-Localization of Online Video
Shi, H., Chen, J., & Hauptmann, A. G. (2017) · ACM International Conference on Multimedia Retrieval (ICMR) · doi:10.1145/3078971.3078996
“In machine learning, a sequence of gradually added training samples [4] is called a curriculum.”
BackgroundSelf-paced Compensatory Deep Boltzmann Machine for Semi-Structured Document Embedding
Li, S., Pan, R., & Yan, J. (2017) · International Joint Conference on Artificial Intelligence · doi:10.24963/ijcai.2017/304
“The curriculum learning [Bengio et al., 2009] provides a learning paradigm where a model is learned by gradually including more complex samples in the training process, which can be used to select the metadata to integrate the text understanding process layerwisely.”
BackgroundMethodGeneral AI Challenge - Round One: Gradual LearningPreprint
Feyereisl, J., Nikl, M., Poliak, M., Stránský, M., & Vlasak, M. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1708.05346 · Cited by 2
“Gradualand guided learning has a number of other be-nefits over other types of learning [Gulcehre et al. , 2016; G¨ulc¸ehre and Bengio, 2016; Bengio et al. , 2009; Pan and Yang, 2010].”
BackgroundMethodInfluentialA parallel computer clusters for convolutional network
Huang, P. S. (2016) · International Computer Conference on Wavelet Active Media Technology and Information Processing (ICCWAMTIP) · doi:10.1109/iccwamtip.2016.8079832
“A sequence of gradually added training samples [5] is called a curriculum.”
BackgroundDistribution-Based Categorization of Classifier Transfer LearningPreprint
Sousa, R., Alexandre, L. A., Santos, J. M., Silva, L. M., & de Sá, J. M. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1712.02159
“This category of TL has also been coined as Never-Ending Learning [40] or Curriculum Learning [9].”
BackgroundNoise self-filtering K-nearest neighbors algorithms
Xia, S., Wang, G., Liur, Y., Liu, Q., & Yu, H. Z. (2017) · IEEE International Conference on Big Data (BigData) · doi:10.1109/bigdata.2017.8258130
“Recently, some efficient learning strategies have also demonstrated the effectiveness, where the easy and difficult data are distinguished and separately processed in a certain order, such as the curriculum learning [4], [30], self-pace learning [31]–[37], and three-way decision [38].”
BackgroundMethodHandling long-term dependencies and rare words in low-resource language modelling
Singh, M. (2017) · Publications of the UdS (Saarland University) · doi:10.22028/d291-26791 · Cited by 1
LSH Microbatches for Stochastic Gradients: Value in Rearrangement.Preprint
Buchnik, E., Cohen, E., Hassidim, A., & Matias, Y. (2018) · arXiv (Cornell University)
Memory Augmented Self-PlayPreprint
Sodhani, S., & Pahuja, V. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1805.11016
“In literature, curriculum learning (Bengio et al., 2009) has been shown to boost the speed of convergence as well as enhance the generalization performance in the domain of supervised learning.”
BackgroundSearch, Abstractions and Learning in Real-Time Strategy Games: A Dissertation Summary
Barriga, N. A. (2019) · Künstliche Intell. · doi:10.1007/s13218-019-00614-0
Michalenko, J., Lan, A., & Baraniuk, R. G. (2017) · Educational Data Mining
Yu, S.-I., Yang, Y., Xu, Z., Xu, S., Meng, D., Mao, Z., Ma, Z., Lin, M., Li, X., Li, H., et al. (2016)
Cold-start recommendation : from Algorithm Portfolios to Job Applicant Matching
Gonard, F. (2018) · theses.fr (ABES) · doi:10.70675/222c3b16z7046z4cf2za432zc050f4bd04e7
Active Collaboration of Classifiers for Visual Tracking
Meshgi, K., & Oba, S. (2018) · InTech eBooks · doi:10.5772/intechopen.74199
“Selecting such queries is addressed in different machine learning domains such as curriculum learning [74] and active learning.”
BackgroundMethodIntegrating Feature and Image Pyramid: A Lung Nodule Detector Learned in Curriculum FashionPreprint
Sun, B., Zhou, Z., Zhang, F., Li, X., & Wang, Y. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1807.08135
Top-Down Tree Structured Text GenerationPreprint
Guo, Q., Qiu, X., Xue, X., & Zhang, Z. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1808.04865
“The first is curriculum learning (Bengio et al., 2009): we constrain the depth and width of the tree to limit the solution space at the beginning of training and gradually relax it.”
BackgroundNeural Task Planning with And-Or Graph RepresentationsPreprint
Chen, T., Chen, R., Nie, L., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1808.09284
“over, because the automatically augmented set includes some difficult samples with uncertain and even incorrect labels, we further incorporate curriculum learning [21], [22] to address this”
BackgroundMethodAn Empirical Comparison of Syllabuses for Curriculum LearningPreprint
Collier, M., & Beel, J. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1809.10789 · Cited by 3
“Additionally explanations for why curriculum learning works have mostly been grounded in learning theory [4,13,2] rather than empirical results.”
BackgroundResultInfluentialGradual Machine Learning for Entity ResolutionPreprint
Hou, B., Chen, Q., Wang, Y., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1810.12125
“We note that there already exist many learning paradigms for a variety of classification tasks, including transfer learning [36], lifelong learning [13], curriculum learning [5], self-paced learning [28] and self-training learning [31] to name a few.”
BackgroundIntra-task Curriculum Learning for Faster Reinforcement Learning in Video Games
du Preez-Wilkinson, N., Gallagher, M. R., & Hu, X. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-03991-2_6
Semantic Segmentation by Integrating Classifiers for Different Difficulty Levels
Matsuzuki, D., & Hotta, K. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-03801-4_53
Task and Instance Quadratic Ordering for Active Online Multitask Learning
Zhao, J. Y., Pang, S., Ardekani, I., Sekiya, Y., & Miyamoto, D. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-04182-3_38
BlockPuzzle - A Challenge in Physical Reasoning and Generalization for Robot LearningPreprint
Zhao, Y., & Ziyin, L. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1812.00091
“eased greatly for these environment. 2.2 Curriculum Learning Of the many methods to guide the agent into the desired behavior, curriculum learning is one of the most general and successful frameworks [BLCW09]. It is also the most intuitive since it is how humans learn most subjects in school. For our environments, we…”
BackgroundInfluentialAn Empirical Evaluation of Sketched SVD and its Application to Leverage Score OrderingPreprint
Chin, H. H., & Liang, P. P. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1812.07903
“We focus on curriculum learning, a research area which aims to determine optimal orderings of training data to improve convergence and performance of neural networks (Bengio et al., 2009).”
BackgroundMethodInfluentialAn Autonomous Model Construction Mechanism in Dynamic Sensor Networks
Ma, L., Bao, Y., & Chen, W. (2018) · IEEE SmartWorld, Ubiquitous Intelligence & Computing, Advanced & Trusted Computing, Scalable Computing & Communications, Cloud & Big Data Computing, Internet of People and Smart City Innovation (SmartWorld/SCALCOM/UIC/ATC/CBDCom/IOP/SCI) · doi:10.1109/smartworld.2018.00285
“Inspired by the idea of curriculum learning [11], which is proposed to leverage the training samples according to a ranking function (also called as curriculum), we design a three-step label refinement strategy, including (1) curriculum generation, (2) curriculum refinement and (3) curriculum teaching, to refine the l…”
BackgroundMethodStudy on Applicability of Curriculum Framework to Feature Selection Algorithms
Kalavala, D., & Bhagvati, C. (2017) · International Conference on Advances in Pattern Recognition · doi:10.1109/icapr.2017.8593126
“[5] have shown in their article that accuracy of the classifier improved over no-curriculum learning.”
BackgroundAl-Qahtani, S., Jamil, A. I., & Kadir, F. A. A. (2018) · AL-TA LIM · doi:10.15548/jt.v25i3.496
Sea Ice Change Detection from Synthetic Aperture Radar Images Based on Self-Paced Boosting Learning
Wang, Q., Gao, F., Dong, J., et al. (2018) · International Workshop on Earth Observation and Remote Sensing Applications · doi:10.1109/eorsa.2018.8598591
“[7] noticed that humans and animals learn much better when the examples are organized in a meaningful order.”
BackgroundUser Feedback and Uncertainty in User Guided Binarization
Westphal, F., Grahn, H., & Lavesson, N. (2018) · IEEE ... International Conference on Data Mining workshops · doi:10.1109/icdmw.2018.00066
“For example, curriculum learning [25] builds the complete training set by initially choosing easy samples and then adding more and more difficult ones to the training set.”
BackgroundSkill Transfer for Mediated Interaction Learning
Fleer, S., & Ritter, H. J. (2018) · IEEE-RAS International Conference on Humanoid Robots (Humanoids) · doi:10.1109/humanoids.2018.8624951
“While most transfer learning schemes are not paying much attention to the fashion in which the source tasks are provided, it can be combined with curriculum learning [9].”
BackgroundTowards Time-Aware Distant Supervision for Relation ExtractionPreprint
Jiang, T., Zhao, S., Liu, J., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1903.03289
Using Structured Input and Modularity for Improved LearningPreprint
Sura, Z., Chen, T., & Sung, H. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1903.12366 · Cited by 1
Curriculum Learning for Tightly Coupled Multiagent Systems
Rockefeller, G., Mannion, P., & Tumer, K. (2019) · Adaptive Agents and Multi-Agents Systems · doi:10.65109/iasz5989
“Curriculum learning (CL) uses TL to improve learning on a target problem by learning first on an easier training task and progressively increasing the difficulty of the training problem, transferring knowledge along the way [3, 4].”
BackgroundAction Recognition From Single Timestamp Supervision in Untrimmed Videos
Moltisanti, D., Fidler, S., & Damen, D. (2019) · Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2019.01015 · Cited by 71
“To decrease noise, we rank frames sampled from all untrimmed videos based on the classifier’s response, and select the most confident frames for training, inspired by curriculum learning [3].”
MethodGenerative Exploration and ExploitationPreprint
Jiang, J., & Lu, Z. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1904.09605
LAW: Learning to Auto Weight.Preprint
Li, Z., Wu, Y., Chen, K., et al. (2019) · arXiv (Cornell University)
Semiempirical Neural Network Models of Controlled Dynamical Systems
Tiumentsev, Y. V., & Вячеславович, Е. М. (2019) · Elsevier eBooks · doi:10.1016/b978-0-12-815254-6.00015-0
Continuous Trade-off Optimization between Fast and Accurate Deep Face DetectorsPreprint
Soviany, P., & Ionescu, R. T. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1811.11582
“Bengio et al. [1] have explored easy-to-hard strategies to train machine learning models, showing that machines can also benefit from learning by gradually adding more difficult examples.”
BackgroundMultiPoseNet: Fast Multi-Person Pose Estimation using Pose Residual NetworkPreprint
Kocabas, M., Karagoz, S., & Akbaş, E. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1807.04067
“We utilized a sort of curriculum learning [51] by sorting annotations based on number of keypoints and bounding box areas.”
MethodLalor, J. P., Wu, H., Munkhdalai, T., et al. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1702.04811
Interactive Learning from Multiple Noisy LabelsPreprint
Vembu, S., & Zilles, S. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1607.06988
“This setting is also known as curriculum learning [1, 2, 3] or self-paced learning [4, 5, 6] in the literature.”
BackgroundLow-Latency Deep Clustering For Speech SeparationPreprint
Wang, S., Naithani, G., & Virtanen, T. I. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1902.07033
“200 (800 ms). We first train the network for 100 frame sequences and then after convergence continue training with 200 frame sequences, known as curriculum learning used in [8] and first introduced in [23]. The idea is to pre-train a network on an easier task first improves learning and generalization. Finally, c) is stu…”
BackgroundSentence-wise Smooth Regularization for Sequence to Sequence LearningPreprint
Gong, C., Tan, X., He, D., & Qin, T. (2018) · AAAI Conference on Artificial Intelligence · doi:10.48550/arxiv.1812.04784 · Cited by 9
“We also compare with a token-level curriculum learning (Bengio et al. 2009; Pentina, Sharmanska, and Lampert 2015) method.”
BackgroundResultMulti-scenario deep learning for multi-speaker source separationPreprint
Zegers, J., & hamme, H. V. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1808.08095
“The networks were trained using curriculum learning [17], i.”
MethodLearning Robust Video Synchronization without AnnotationsPreprint
Wieschollek, P., Freeman, I., & Lensch, H. P. A. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1610.05985 · Cited by 2
Adversarial Self-Paced Learning for Mixture Models of Hawkes ProcessesPreprint
Luo, D., Xu, H., & Carin, L. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.08397
“Then, based on MLE with an adversarial self-paced (Bengio et al., 2009; Kumar et al., 2010) regularizer, we use these candidate sequences to learn the target model and select “easy” sequences for the next iteration.”
Growing Action SpacesPreprint
Farquhar, G., Gustafson, L., Lin, Z., Whiteson, S., Usunier, N., & Synnaeve, G. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1906.12266 · Cited by 9
“Such curricula can drastically improve sample efficiency (Bengio et al., 2009).”
BackgroundKatzmann, A., Mühlberg, A., Sühling, M., Nörenberg, D., & Groß, H. (2019) · IEEE International Symposium on Biomedical Imaging (ISBI) · doi:10.1109/isbi.2019.8759505
“The process of metamemory importance sampling might seem similar to the idea of curriculum learning [18].”
BackgroundGradual Sampling Gate for Bidirectional Knowledge Distillation
Minami, S., Yamashita, T., & Fujiyoshi, H. (2019) · Machine Vision Applications · doi:10.23919/mva.2019.8757888
“Distillation has recently been connected to curriculum learning [3] in information learning theory [11].”
BackgroundScalable Robust Models Under Adversarial Data Corruption
Zhang, X. (2019) · VTechWorks (Virginia Tech)
Quantifying Similarity between Relations with Fact DistributionPreprint
Chen, W., Zhu, H., Han, X., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1907.08937
“This can be also viewed as a process of curriculum learning(Bengio et al., 2009), the data fed to the model gradually changes from simple negative triples to hard ones.”
BackgroundActive collaborative ensemble tracking
Meshgi, K., Mirzaei, M. S., Oba, S., & Ishii, S. (2017) · IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS) · doi:10.1109/avss.2017.8078534 · Cited by 1
“Different studies have tried to provide this small clean subset by different approaches: pruning outliers [12] and hard-to-learn samples [1], learning easy-to-classify examples first (as known as the Curriculum learning) [8], treating samples as noisy observations [14], defining a training value for each sample by tre…”
BackgroundRobby is Not a Robber (anymore): On the Use of Institutions for Learning Normative BehaviorPreprint
Tomic, S., Pecora, F., & Saffiotti, A. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.02138
“Another unexplored possibility would be to realize curriculum learning [35] by gradually increasing the difficulty of the problem by changing the grounding.”
BackgroundSequentially Aggregated Convolutional NetworksPreprint
Huang, Y., Wu, R., Ou, P., et al. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1811.10798
“The progressive growing mechanism can be interpreted as a form of curriculum learning [2] that it enables intermediate representations to be hierarchically explored.”
BackgroundTransductive Auxiliary Task Self-Training for Neural Multi-Task Models
Bjerva, J., Kann, K., & Augenstein, I. (2019) · Workshop on Deep Learning Approaches for Low-Resource NLP (DeepLo) · doi:10.18653/v1/d19-6128
“Another option is curriculum learning, where selection is based on learning difficulty, increasing the difficulty during learning (Bengio et al., 2009).”
BackgroundEmbedding Big Data in Graph Convolutional Networks
Palaiopanos, G., Stalidis, P., Vretos, N., et al. (2019) · IEEE International Conference on Engineering, Technology and Innovation (ICE/ITMC) · doi:10.1109/ice.2019.8792632
“As analyzed in [16], when the examples are not randomly presented but organized in a meaningful order which illustrates concepts in a gradually increasing quantity and complexity, the learning procedure is expected to improve.”
BackgroundMethodVariational Autoencoded Regression: High Dimensional Regression of Visual Data on Complex ManifoldPreprint
Yoo, Y., Yun, S., Chang, H. J., et al. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1908.04015
“Mini-Batch Training: The work in [7] reports that the composition of the mini-batch is critical when using variants of stochastic gradient descent methods [23, 10] to train the parameters.”
MethodCD-ABM: Curriculum Design with Attention Branch Model for Person Re-identification
Chen, J., Qian, J., Zhu, X., et al. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-29894-4_50
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.