Cited by 740Machine Learning and AlgorithmsMachine Learning and Data ClassificationDomain Adaptation and Few-Shot Learning

The DOI matches its Crossref record; 221 of the 223 references it cites were found in a registry or an index.

Multiple instance learning: A survey of problem characteristics and applications

Open access

Carbonneau, Marc-André, Cheplygina, Veronika, Granger, Eric, Gagnon, Ghyslain·2018·Pattern Recognition

DOI 10.1016/j.patcog.2017.10.009

Open this paper in CoChat — free account Open PDF

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

Integrity signals

Registry

Crossref · 10.1016/j.patcog.2017.10.009

Cites

221 of 221 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

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?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

ImageNet Large Scale Visual Recognition Challenge
Russakovsky, O., Deng, J., Su, H., Krause, J., Satheesh, S., Ma, S., Huang, Z., Karpathy, A., Khosla, A., Bernstein, M., et al. (2015) · International Journal of Computer Vision · doi:10.1007/s11263-015-0816-y · Cited by 39,380
SMOTE synthetic minority over-sampling technique
Chawla, N. V., Bowyer, K. W., Hall, L. O., & Kegelmeyer, W. P. (2002) · Journal of Artificial Intelligence Research · doi:10.1613/jair.953 · Cited by 32,865 · cited 2×
“There are external methods like SMOTE [193] and RUSBoost [194] that necessitate accurate labels to perform over or under sampling.”
Method
Greedy function approximation: A gradient boosting machine
Friedman, J. H. (2001) · The Annals of Statistics · doi:10.1214/aos/1013203451 · Cited by 30,731 · cited 1×
“The method is essentially the same as gradient boosting [143] except that the loss function is based on bag classification error.”
Method
The pascal visual object classes (VOC) challenge
Everingham, M., Van Gool, L., Williams, C. K. I., Winn, J., & Zisserman, A. (2010) · International Journal of Computer Vision · doi:10.1007/s11263-009-0275-4 · Cited by 19,599
The WEKA data mining software: an update
Hall, M., Frank, E., Holmes, G., Pfahringer, B., Reutemann, P., & Witten, I. H. (2009) · ACM SIGKDD Explorations Newsletter · doi:10.1145/1656274.1656278 · Cited by 17,750
Representation Learning: A Review and New Perspectives
Bengio, Y., Courville, A., & Vincent, P. (2013) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2013.50 · Cited by 13,001 · cited 1×
“Recently, methods for representation learning [160] have gained in popularity because they usually yield a high level of accuracy.”
Method
Statistical Comparisons of Classifiers over Multiple Data Sets
Demšar, J. (2006) · doi:10.5555/1248547.1248548 · Cited by 12,051
Cross validatory choice and assessment of statistical predictions
Stone, M. (1974) · Journal of the Royal Statistical Society Series B: Statistical Methodology · doi:10.1111/j.2517-6161.1974.tb00994.x · Cited by 10,881 · cited 1×
“The experiments were conducted using a nested cross-fold validation protocol [149].”
Method
Beyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories
Lazebnik, S., Schmid, C., & Ponce, J. (2006) · 2006 IEEE Computer Society Conference on Computer Vision and Pattern Recognition - Volume 2 (CVPR'06) · doi:10.1109/cvpr.2006.68 · Cited by 7,916 · cited 1×
“In BoW models for computer vision, this can be achieved by dividing the images [102, 103] or videos [80] into different spatial and/or temporal zones.”
The Earth Mover's Distance as a Metric for Image Retrieval
Rubner, Y., Tomasi, C., & Guibas, L. (2000) · International Journal of Computer Vision · doi:10.1023/a:1026543900054 · Cited by 5,547

Show all 223 references

Research works printed but not linked: 1

Babenko, 2011, Robust object tracking with online multiple instance learning, IEEE Trans. Pattern Anal. Mach. Intell., 33, 1619, 10.1109/TPAMI.2010.226 · DOI · Paper page

Sorted by citations

Items 301–400 of 734

A deep learning model to triage and predict adenocarcinoma on pancreas cytology whole slide imaging
Sohn, A., Miller, D. L., Ribeiro, E., Shankar, N., Ali, S. Z., Hruban, R. H., & Baras, A. S. (2023) · Scientific Reports · doi:10.1038/s41598-023-42045-w · Cited by 10
“The standard operating procedure in deep learning which involve the classification of ultra-high resolution images (e.g. satellite images in remote sensing as another example) use the multiple instance learning (MIL) paradigm 24 .”
Method
A data stream-based approach for anomaly detection in surveillance videos
Aydoğdu, Ö., & Ekіncі, M. (2024) · Multimedia Tools and Applications · doi:10.1007/s11042-023-17861-x · Cited by 9
“A multiple instance learning (MIL) based method [9] is employed for weak labelling of the current sub video instance.”
Weakly Supervised AUC Optimization: A Unified Partial AUC Approach
Xie, Z., Liu, Y., He, H.-Y., Li, M., & Zhou, Z. (2024) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2024.3357814 · Cited by 9
“The framework provides a uniﬁed view of multiple WSL scenarios including (1) noisy label learning, (2) positive-unlabeled learning, (3) multi-instance learning, and (4) semi-supervised learning with or without label noise.”
Background
Robust fuzzy clustering for multiple instance regression
Trabelsi, M., & Frigui, H. (2019) · Pattern Recognition · doi:10.1016/j.patcog.2019.01.030 · Cited by 9
Weakly Supervised Attention Networks for Fine-Grained Opinion Mining and Public Health
Karamanolakis, G., Hsu, D., & Gravano, L. (2019) · Workshop on Noisy User-generated Text (W-NUT) · doi:10.18653/v1/d19-5501 · Cited by 11
“The choice of aggregation function has a substantial impact on the performance of MIL-based models and should depend on the specific assumptions about the relationship between bags and instances (Carbonneau et al., 2018).”
Background
Training deep retrieval models with noisy datasets: Bag exponential loss
Martínez-Cortés, T., González-Díaz, I., & Díaz-de-María, F. (2020) · Pattern Recognition · doi:10.1016/j.patcog.2020.107811 · Cited by 9
Multiple Instance Learning for ECG Risk StratificationPreprint
Shanmugam, D., Blalock, D. W., & Guttag, J. V. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1812.00475 · Cited by 9
“A wealth of MIL research has occurred since the field’s inception in 1998 (Maron and Lozano-Pérez, 1998) and we direct interested readers to the survey conducted by (Carbonneau et al., 2018).”
Background
A Graph Convolutional Multiple Instance Learning on a Hypersphere Manifold Approach for Diagnosing Chronic Obstructive Pulmonary Disease in CT Images
Chen, L., Feng, Q., Yin, X. P., Min, X., Shi, L., Yang, D., Chen, Y., Zhang, D., & Zhu, W. (2022) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2022.3209410 · Cited by 8
“Multiple instance learning (MIL) is a kind of weakly supervised learning method that arranges training instances in sets [9].”
Background
Protein-ligand binding affinity prediction using multi-instance learning with docking structures
Kim, H.-J., Shim, H., Ranganath, A., He, S., Stevenson, G. A., & Allen, J. (2025) · Frontiers in Pharmacology · doi:10.3389/fphar.2024.1518875 · Cited by 9
Multimodal medical imaging AI for breast cancer diagnosis: A comprehensive review
Wei, T.-R., & Yan, Y. (2025) · Intelligent Oncology · doi:10.1016/j.intonc.2025.12.005 · Cited by 10
Psychophysiological Arousal in Young Children Who Stutter
Sharma, H., Xiao, Y., Tumanova, V., & Salekin, A. (2022) · Proceedings of the ACM on Interactive Mobile Wearable and Ubiquitous Technologies · doi:10.1145/3550326 · Cited by 8
“There are two variety of MIL approaches in literature [14, 35]: SHARMA et al.”
Abductive subconcept learning
Han, Z., Cai, L.-W., Dai, W.-Z., Huang, Y., Wei, B., Wang, W., & Yin, Y. (2023) · Science China Information Sciences · doi:10.1007/s11432-020-3569-0 · Cited by 9
ProMIL: Probabilistic Multiple Instance Learning for Medical Imaging
Struski, Ł., Rymarczyk, D., Lewicki, A., Sabiniewicz, R., Tabor, J., & Zieliński, B. (2023) · Frontiers in artificial intelligence and applications · doi:10.3233/faia230518 · Cited by 9
Whole slide image representation in bone marrow cytology
Mu, Y., Tizhoosh, H. R., Dehkharghanian, T., & Campbell, C. J. V. (2023) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2023.107530 · Cited by 9
Multiple instance learning for medical image classification based on instance importance
Struski, Ł., Janusz, S., Tabor, J., Markiewicz, M., & Lewicki, A. (2023) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2023.105874 · Cited by 10
Learning to Find Common Objects Across Few Image Collections
Shaban, A., Rahimi, A. M., Bansal, S., Gould, S. J., Boots, B., & Hartley, R. I. (2019) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00522 · Cited by 8
“MI-SVM and mi-SVM [2] are two popular methods for MIL, and have been widely adapted for many weakly supervised computer vision problems, achieving state-of-the-art results in many different applications [7, 13].”
Method
U-Net adaptation for multiple instance learning
Kots, M. V., & Chukanov, V. S. (2019) · Journal of Physics Conference Series · doi:10.1088/1742-6596/1236/1/012061 · Cited by 8
A novel method for clinical risk prediction with low-quality data
Wang, Z., Poon, J., Wang, S., Sun, S., & Poon, S. K.-S. (2021) · Artificial Intelligence in Medicine · doi:10.1016/j.artmed.2021.102052 · Cited by 8
Explainable Fraud Detection for Few Labeled Time Series Data
Xiao, Z., & Jiao, J. (2021) · Security and Communication Networks · doi:10.1155/2021/9941464 · Cited by 8
A QuadTree Image Representation for Computational Pathology
Jewsbury, R., Bhalerao, A. H., & Rajpoot, N. (2021) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw54120.2021.00078 · Cited by 10
“Instance-based: Here f classifies each instance individually, assigning a class label for each instance and then α combines these predictions together to generate an overall prediction for the bag using operations such as mean, maximum etc [4].”
Method
Nested Multiple Instance Learning with Attention Mechanisms
Fuster, S., Eftestøl, T., & Engan, K. (2022) · International Conference on Machine Learning and Applications · doi:10.1109/icmla55696.2022.00038 · Cited by 11
“Several applications have been developed that use MIL for overcoming the lack of annotated ROIs in an image [14, 15].”
Method
The Weak Supervision Landscape
Poyiadzi, R., Bacaicoa-Barber, D., Cid‐Sueiro, J., Perelló-Nieto, M., Flach, P. A., & Santos‐Rodríguez, R. (2022) · IEEE International Conference on Pervasive Computing and Communications Workshops (PerCom Workshops) · doi:10.1109/percomworkshops53856.2022.9767420 · Cited by 8
IDA-MIL: Classification of Glomerular with Spike-like Projections via Multiple Instance Learning with Instance-level Data Augmentation
Wu, X., Chen, Y., Li, X., Liu, X., Liu, Y., Wu, Y., Li, M., Zhou, X., & Wang, C. (2022) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2022.107106 · Cited by 8
Contribution of topographic features and categorization uncertainty for a tree species classification in the boreal biome of Northern Ontario
Pittman, R., & Baoxin Hu, G. (2023) · GIScience & Remote Sensing · doi:10.1080/15481603.2023.2214994 · Cited by 8
“The issue of noisy labels can be addressed by hierarchical constrained energy minimum (HCEM) methods (Bing et al. 2020) or by weakly supervised multiple instance learning (MIL) (Carbonneau et al. 2018).”
Background
The semiproximal SVM approach for multiple instance learning: a kernel-based computational study
Avolio, M., & Fuduli, A. (2023) · Optimization Letters · doi:10.1007/s11590-023-02022-8 · Cited by 8
“For solving a MIL problem, in the literature there exist various approaches [1, 9], depending on the space where the classification process is performed.”
Instance-level loss based multiple-instance learning framework for acoustic scene classification
Choi, W.-G., Chang, J., Yang, J., & Moon, H.-G. (2023) · Applied Acoustics · doi:10.1016/j.apacoust.2023.109757 · Cited by 8
“As the proposed instance-level loss-based MIL framework overcame the underestimation problem, the performance outperformed the SupL-based ASC system.”
Method
A multi-instance support vector machine with incomplete data for clinical outcome prediction of COVID-19
Brand, L., Baker, L. Z., & Wang, H. (2021) · ACM International Conference on Bioinformatics, Computational Biology, and Health Informatics (ACM-BCB) · doi:10.1145/3459930.3469552 · Cited by 8
“Our approach relies on combining techniques from multi-instance learning [6, 15, 18–20], as well as matrix completion [5], to handle missing data across an entire patient cohort.”
Method
Towards Visual Anomaly Detection in Domains with Limited Amount of Labeled Data
Štepec, D., & Skočaj, D. (2020) · Računalniška obdelava slik in njena uporaba v Sloveniji (ROSUS) · doi:10.18690/978-961-286-337-1.1 · Cited by 6
In the Wild Video Violence Detection: An Unsupervised Domain Adaptation Approach
Ciampi, L., Santiago, C., Falchi, F., Gennaro, C., & Amato, G. M. (2024) · SN Computer Science · doi:10.1007/s42979-024-03126-3 · Cited by 7
“MIL is a type of weakly supervised learning in which training instances are organized into groups, referred to as "bags", and a single label is assigned to the entire bag [46]; in our context, bags are represented by the trimmed videos while instances are the frames composing the clips themselves.”
Multiple Instance Learning: State-of-the-Art in Cancer Analysis, Applications & Future Potential
Mozumder, M. A. I., Sumon, R. I., Uddin, S. M. I., Armand, T. P. T., & Kim, H. (2024) · International Conference on Electrical, Computer and Energy Technologies (ICECET) · doi:10.1109/icecet61485.2024.10698347 · Cited by 7
“MIL has applications in various domains such as image analysis, drug discovery, and anomaly detection, where data is naturally organized into groups or collections [1].”
Background
CoLM: Contrastive learning and multiple instance learning network for lung cancer classification of surgical options based on frozen pathological images
Zhao, L., Zhao, W., Qiu, L., Jiang, M., Qian, L., Ting, H.-N., Fu, X., Zhang, P., Han, Y., & Zhao, J. (2024) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2024.107097 · Cited by 7
Diagnostic classification in toxicologic pathology using attention-guided weak supervision and whole slide image features: a pilot study in rat livers
Zehnder, P., Feng, J., Nguyen, T., Shen, P., Sullivan, R., Fuji, R. N., & Hu, F. (2025) · Scientific Reports · doi:10.1038/s41598-025-86615-6 · Cited by 7
Smooth Attention for Deep Multiple Instance Learning: Application to CT Intracranial Hemorrhage Detection
Wu, Y., Castro-Macías, F. M., Morales-Álvarez, P., Molina, R., & Katsaggelos, A. K. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-43904-9_32 · Cited by 8
A maximum-margin multisphere approach for binary Multiple Instance Learning
Astorino, A., Avolio, M., & Fuduli, A. (2021) · European Journal of Operational Research · doi:10.1016/j.ejor.2021.11.022 · Cited by 7
A fisher score-based multi-instance learning method assisted by mixture of factor analysis
Huang, S., Liu, Z., Jin, W., & Mu, Y. (2022) · Neurocomputing · doi:10.1016/j.neucom.2022.07.075 · Cited by 7
Using multiple instance learning for explainable solar flare prediction
Huwyler, C., & Melchior, M. (2022) · Astronomy and Computing · doi:10.1016/j.ascom.2022.100668 · Cited by 7
“relationship between individual instances and the overall bag label is made explicit by modeling how instance-level information is aggregated to bag-level information (Carbonneau et al., 2018; Amores, 2013; Foulds and Frank, 2010).”
Method
Detecting Depression from Social Media Data as a Multiple-Instance Learning Task
Mann, P., Matsushima, E. H., & Paes, A. (2022) · International Conference on Affective Computing and Intelligent Interaction and workshops · doi:10.1109/acii55700.2022.9953811 · Cited by 8
“In this way, as labels are originally tied to users, not posts, fabricating individual instance labels is misleading; to avoid that, we must relax the bag label assumption or let the model automatically learn it [5], as we do here.”
Background
Robust handcrafted features for music genre classification
Muniz, V. H. D. S., & Filho, J. B. D. O. E. S. (2022) · Neural Computing and Applications · doi:10.1007/s00521-022-08069-5 · Cited by 8
Multi-Objective Multi-Instance Learning: A New Approach to Machine Learning for eSports
Birant, K. U., & Birant, D. (2022) · Entropy · doi:10.3390/e25010028 · Cited by 7
“Multi-instance learning (MIL) [20] is a special type of machine learning (ML) where multiple training samples are assigned into bags and only one class label is assigned for all the samples in a bag.”
Background
Prediction of drug-induced hepatotoxicity based on histopathological whole slide images
Su, R., He, H., Sun, C., Wang, X., & Liu, X. (2023) · Methods · doi:10.1016/j.ymeth.2023.01.005 · Cited by 7
Weakly supervised identification of microscopic human breast cancer-related optical signatures from normal-appearing breast tissue
Shi, J., Tu, H., Park, J., Marjanović, M., Higham, A. M., Luckey, N. N., Cradock, K. A., Liu, Z., & Boppart, S. A. (2023) · Biomedical Optics Express · doi:10.1364/boe.480687 · Cited by 7
Multi-Instance Learning Using Transformer for Obscured Weak Label Problem in PPG-Based Blood Pressure Estimation
Hu, Q., Wang, D., & Yang, C. (2023) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2023.3324675 · Cited by 7
“More importantly, the proposed BP-MIL-Trans exhibited the best performance compared to other baseline models over four subsets. ground-noise labeling has been underscored in a review paper [37].”
Background
Towards Hierarchical Regional Transformer-based Multiple Instance Learning
Cersovsky, J., Mohammadi, S., Kainmueller, D., & Hoehne, J. (2023) · IEEE International Conference on Computer Vision Workshops · doi:10.1109/iccvw60793.2023.00427 · Cited by 7
“However, the incomplete, sparse and often noisy nature of random patch-sampling remains a challenging hurdle and a limiting factor to predictive performance [4, 12].”
Background
Logistic Regression based Feature Selection and Two-Stage Detection for EEG based Motor Imagery Classification
Mada, U. G., Wijaya, A., Adji, T. B., Mada, U. G., Setiawan, N. A., & Mada, U. G. (2020) · International journal of intelligent engineering and systems · doi:10.22266/ijies2021.0228.14 · Cited by 7
“This detection, so called multi-instance learning (MIL) with instance-level prediction [43].”
BackgroundMethod
AMI-Net+: A Novel Multi-Instance Neural Network for Medical Diagnosis from Incomplete and Imbalanced DataPreprint
Wang, Z., Poon, J., & Poon, S. K.-S. (2019) · arXiv (Cornell University) · doi:10.48550/arxiv.1907.01734 · Cited by 7
“In addition, when calculating the bag representation, the correlations of instances are essential to explore [30] and self-attention mechanism [23] shows its superiority in this respect.”
Background
Learning to Select Cuts for Efficient Mixed-Integer ProgrammingPreprint
Huang, Z., Wang, K., Liu, F., Zhen, H.-L., Zhang, W., Yuan, M., Hao, J., Yu, Y. H., & Wang, Y.-X. (2021) · arXiv (Cornell University) · doi:10.48550/arxiv.2105.13645 · Cited by 7
TA-NET: Empowering Highly Efficient Traffic Anomaly Detection Through Multi-Head Local Self-Attention and Adaptive Hierarchical Feature Reconstruction
Chen, J., Pu, J., Yin, B., Zhang, R., & Wu, J. J. (2024) · IEEE Transactions on Intelligent Transportation Systems · doi:10.1109/tits.2024.3365820 · Cited by 6
Dual-Attention Multiple Instance Learning Framework for Pathology Whole-Slide Image Classification
Liu, D., Li, C., Hu, X., & Hu, B. (2024) · Electronics · doi:10.3390/electronics13224445 · Cited by 6
Dual-path neural network extracts tumor microenvironment information from whole slide images to predict molecular typing and prognosis of Glioma
Ning, Z., Yang, B., Wang, Y., Shi, Z., Yu, J., & Wu, G. (2025) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2024.108580 · Cited by 6
Lung Cancer Screening Classification by Sequential Multi-Instance Learning (SMILE) Framework With Multiple CT Scans
Zhao, W., Fu, Y., Shen, Y., Ma, J., Zhao, L., Fu, X., Zhang, P.-M., & Zhao, J. (2025) · IEEE Transactions on Medical Imaging · doi:10.1109/tmi.2025.3559143 · Cited by 8
“In medical image analysis field, MIL was used commonly in the situation without enough instance-level labels, especially in pathological images [22], [23].”
Classification of Whole-Slide Pathology Images Based on State Space Models and Graph Neural Networks
Ding, F., Cai, C., Li, J., Liu, M., Jiao, Y., Wu, Z., & Xu, J. (2025) · Electronics · doi:10.3390/electronics14102056 · Cited by 6
Dynamic hypergraph representation for bone metastasis analysis
Chen, Y., Li, J., Zhu, L., Yang, X., Guan, T., Shi, H., Yonghong, H., & Han, A. (2025) · Computer Methods and Programs in Biomedicine · doi:10.1016/j.cmpb.2025.108966 · Cited by 7
Risk assessment of thyroid nodules with a multi-instance convolutional neural network
Da, Y., Song, T., Yu, Y., Zhang, H., Gao, F., Wang, Z., & Wang, J. (2025) · Frontiers in Oncology · doi:10.3389/fonc.2025.1608963 · Cited by 6
“In this context, MIL is employed to analyze all images of a patient collectively, synthesizing the information from each instance to yield a more accurate and holistic risk assessment (30, 31).”
A weakly supervised end-to-end framework for semantic segmentation of cancerous area in whole slide image
Feng, Y., Hafiane, A., & Laurent, H. (2024) · Pattern Analysis and Applications · doi:10.1007/s10044-024-01251-6 · Cited by 6
“Several applications of MIL have been found in the cancerous diagnosis of WSI [11–13], they are based on an assumption that there must be at least one patch cropped from a cancerous WSI that has cancerous cells, while there is none patch cropped from a normal WSI that contains cancerous cells.”
Herb-Target Interaction Prediction by Multiinstance Learning
Zhu, Y., Ren, L.-R., Sun, R., Wang, J., & Yu, G. (2024) · IEEE Transactions on Artificial Intelligence · doi:10.1109/tai.2024.3515932 · Cited by 6
“Not all instances within a bag are equally informative [38].”
A simple yet effective approach for log based critical errors prediction
Lopez, M., Beurton‐Aimar, M., Diallo, G., & Maabout, S. (2022) · Computers in Industry · doi:10.1016/j.compind.2021.103605 · Cited by 6
Identification lymph node metastasis in esophageal squamous cell carcinoma using whole slide images and a hybrid network of multiple instance and transfer learning
Kang, H., Yang, M., Zhang, F., Xu, H., Ren, S., Li, J., Chen, D., Wang, F., Li, D., & Chen, X. (2023) · Biomedical Signal Processing and Control · doi:10.1016/j.bspc.2023.104577 · Cited by 6
Counting is almost all you need
Akerman, O., Isakov, H., Levi, R., Psevkin, V., & Louzoun, Y. (2023) · Frontiers in Immunology · doi:10.3389/fimmu.2022.1031011 · Cited by 6
A linear primal–dual multi-instance SVM for big data classifications
Brand, L., Seo, H., Baker, L. Z., Ellefsen, C., Sargent, J., & Wang, H. (2023) · Knowledge and Information Systems · doi:10.1007/s10115-023-01961-z · Cited by 6
Multi-instance learning with application to the profiling of multi-victim homicides
Delgado, R., & Sánchez-Delgado, H. (2023) · Expert Systems with Applications · doi:10.1016/j.eswa.2023.121593 · Cited by 6
MCPNET: Development of an interpretable deep learning model based on multiple conformations of the compound for predicting developmental toxicity
Cao, C., Wang, H., Yang, J., Chen, Q., Guo, Y.-M., & Chen, J. (2024) · Computers in Biology and Medicine · doi:10.1016/j.compbiomed.2024.108037 · Cited by 6
XFibrosis: Explicit Vessel-Fiber Modeling for Fibrosis Staging from Liver Pathology Images
Yin, C., Liu, S., Lyu, F., Lu, J., Darkner, S., Wong, V. W., & Yuen, P. C. (2024) · IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings - IEEE Computer Society Conference on Computer Vision and Pattern Recognition/Proceedings · doi:10.1109/cvpr52733.2024.01072 · Cited by 6
“Multiple-instance learning (MIL) [8, 21, 29] demonstrates a robust capability in handling whole slide images where only slide-level labels are available.”
Background
Multiple-instance learning via multiple-point concept based instance selection
Yuan, L., Xu, G., Zhao, L., Wen, X., & Xu, H. (2020) · International Journal of Machine Learning and Cybernetics · doi:10.1007/s13042-020-01105-7 · Cited by 6
Cohort selection for clinical trials using multiple instance learning
Dai, H.-J., Wang, F.-D., Chen, C., Su, C.-H., Wu, C., & Jonnagaddala, J. (2020) · Journal of Biomedical Informatics · doi:10.1016/j.jbi.2020.103438 · Cited by 6
Certainty Pooling for Multiple Instance Learning
Gildenblat, J., Ben-Shaul, I., Lapp, Z., & Klaiman, E. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-68763-2_11 · Cited by 6
svMIL: predicting the pathogenic effect of TAD boundary-disrupting somatic structural variants through multiple instance learning
Nieboer, M. M., & de Ridder, J. (2020) · Bioinformatics · doi:10.1093/bioinformatics/btaa802 · Cited by 6
ProtoMIL: Multiple Instance Learning with Prototypical Parts for Fine-Grained InterpretabilityPreprint
Rymarczyk, D., Kaczyńska, A., Kraus, J., Pardyl, A., & Zieliński, B. (2021) · arXiv (Cornell University) · Cited by 6
Reducing Label Dependency in Human Activity Recognition with Wearables: From Supervised Learning to Novel Weakly Self-Supervised Approaches
Sheng, T., & Huber, M. (2025) · Sensors · doi:10.3390/s25134032 · Cited by 5
“Common weakly supervised strategies include multiple instance learning using group-level labels [28], label propagation from a small labeled subset [29], and contrastive methods with pairwise constraints [30,31] (e.g., must-link, cannot-link).”
RefCap: Zero-shot Video Corpus Moment Retrieval Based on Refined Dense Video Captioning
Pan, Y. F., Zhang, Y., Kampffmeyer, M., & Zhao, X. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10888164 · Cited by 5
“…k ∈ K ih is computed as: To measure the similarity between two keyword sets K T and K ih , we define the Max Mean operation as: where the inside max follows the Multiple Instance Learning [16] paradigm, and the outer average operation measures the matching degree of K ih to the overall query T .”
TAIL-MIL: Time-Aware and Instance-Learnable Multiple Instance Learning for Multivariate Time Series Anomaly Detection
Jang, J., & Kwon, H.-Y. (2025) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v39i17.33933 · Cited by 5
“It is effective when a single bag is composed of multiple instances, and individual labels for each instance are not provided (Javed et al. 2022; Carbonneau et al. 2018).”
Weakly Supervised Deep Learning for Monitoring Sleep Apnea Severity Using Coarse-Grained Labels
Zan, X., Wang, D., Song, C., Liu, F., Xian, X., & Berry, R. (2025) · IEEE Transactions on Automation Science and Engineering · doi:10.1109/tase.2025.3566682 · Cited by 5
“While this assumption has proven e ﬀ ective in early MIL works applied in drug activity prediction, it is not guaranteed to hold in many other domains [36].”
Learnable prototype-guided multiple instance learning for detecting tertiary lymphoid structures in multi-cancer whole-slide pathological images
Xia, P., Chen, D., An, H., Lim, K. S., & Yang, X. (2025) · Medical Image Analysis · doi:10.1016/j.media.2025.103652 · Cited by 5
Imbalanced Histopathology Image Classification Using Deep Feature Graph Attention Network
Cong, C., Yang, Y., Liu, S., Pagnucco, M., & Song, Y. (2022) · IEEE International Symposium on Biomedical Imaging (ISBI) · doi:10.1109/isbi52829.2022.9761619 · Cited by 5
“Second, to further reduce the impact of the difference in the number of samples between classes, inspired by the standard multiple instance learning assumption [10], we introduce a minority preferred inference procedure.”
Method
On the Relevance of Temporal Features for Medical Ultrasound Video Recognition
Smith, D. H., Lineberger, J. P., & Baker, G. H. (2023) · Lecture notes in computer science · doi:10.1007/978-3-031-43895-0_70 · Cited by 5
“Instances within a bag are assumed to be unordered and, conditional on the bag label, independent from one another [2].”
Background
Lead optimization
Ashenden, S. K. (2021) · Elsevier eBooks · doi:10.1016/b978-0-12-820045-2.00007-6 · Cited by 6
A deep tensor-based approach for automatic depression recognition from speech utterances
Pandey, S., Shekhawat, H. S., Prasanna, S. R. M., Bhasin, S., & Jasuja, R. (2022) · PLoS ONE · doi:10.1371/journal.pone.0272659 · Cited by 6
Domain Adaptive Multiple Instance Learning for Instance-Level Prediction of Pathological Images
Takahama, S., Kurose, Y., Mukuta, Y., Abé, H., Yoshizawa, A., Ushiku, T., Fukayama, M., Kitagawa, M., Kitsuregawa, M., & Harada, T. (2023) · IEEE International Symposium on Biomedical Imaging (ISBI) · doi:10.1109/isbi53787.2023.10230546 · Cited by 5
Vague-Segment Technique: Automatic Computation of Tumor Stroma Ratio for Breast Cancer on Whole Slides
Xinsen, L., Yang, K., Bingzhi, C., Xiuhong, C., Xinling, L., Xu, X., Chen, J., Ming, T., Pengtao, L., Zheng, X., et al. (2023) · IEEE Journal of Biomedical and Health Informatics · doi:10.1109/jbhi.2023.3341101 · Cited by 5
“…recent dramatic increasing interests in artiﬁcial intelligence techniques, diagnostic analysis researchers have relied on Multi-Instance Learning [12] and Semantic Segmentation [13] to create artiﬁcial intelligence algorithms for the study of breast cancer and TSR. Speciﬁcally, several…”
Overview of the Potentials of Multiple Instance Learning in Cancer Diagnosis: Applications, Challenges, and Future Directions
Armand, T. P. T., Bhattacharjee, S., & Kim, H. (2024) · International Conference on Advanced Communications Technology (ICACT) · doi:10.23919/icact60172.2024.10471995 · Cited by 5
“MIL models can effectively handle this variability by considering the collective information of instances within a bag, enabling the identification of common patterns or features associated with the presence or absence of cancer [34].”
Background
Exploiting Instance-level Relationships in Weakly Supervised Text-to-Video Retrieval
Yin, S., Zhao, S., Wang, H., Xu, T., & Chen, E. (2024) · ACM Transactions on Multimedia Computing Communications and Applications · doi:10.1145/3663571 · Cited by 5
“Most existing MIL methods, including the seminal work for PRVR [15], make the independent and identically distributed (i.i.d.) assumption [7], assuming that positive and negative instances are independently sampled from positive and negative distributions.”
FETR: A Weakly Self-Supervised Approach for Fetal Ultrasound Anatomical Detection
Khan, U., Nawaz, U., Khan, M., Saddik, A. E., & Gueaieb, W. (2024) · IEEE International Symposium on Medical Measurements and Applications (MeMeA) · doi:10.1109/memea60663.2024.10596798 · Cited by 5
“In this work, we introduce a new approach that integrates the Multiple Instance Learning (MIL) [8] with the Detection Transformer (DETR) [9] model.”
Method
TempTrans-MIL: A General Approach to Enhancing Multimodal Tactile-Driven Robotic Manipulation Classification Tasks
Wang, J., Ouyang, W., Fang, S., Zhang, Y., Wu, X., & Yi, Z. (2025) · IEEE/ASME Transactions on Mechatronics · doi:10.1109/tmech.2025.3546938 · Cited by 5
“2) Multiple Instance Learning in Time Series Modeling: MIL ,asitidentiﬁespatternsacrossmultipleinstances withouttheneedfordetailedinstance-levellabels[28].”
Customer Segmentation Using Multiple Instance Clustering and Purchasing Behaviors
Fuentes, I., Nápoles, G., Arco, L., & Vanhoof, K. (2018) · Lecture notes in computer science · doi:10.1007/978-3-030-01132-1_22 · Cited by 5
Semi-supervised Learning for Medical Image Segmentation
Kots, M. V., Pozigun, M., Konstantinov, A. V., & Chukanov, V. S. (2021) · Smart innovation, systems and technologies · doi:10.1007/978-981-33-6632-9_21 · Cited by 5
Correlation-Based Discovery of Disease Patterns for Syndromic Surveillance
Rapp, M. T., Kulessa, M., Mencía, E. L., & Fürnkranz, J. (2022) · Frontiers in Big Data · doi:10.3389/fdata.2021.784159 · Cited by 5
“The problem of learning from assignments of target variables to sets of instances, rather than individual instances, is known as multiple instance learning (Carbonneau et al., 2018).”
Background
SPECHT: Self-tuning Plausibility based object detection Enables quantification of Conflict in Heterogeneous multi-scale microscopy
Cardoen, B., Wong, T., Alan, P., Lee, S., Matsubara, J. A., Nabi, I. R., & Hamarneh, G. (2022) · PLoS ONE · doi:10.1371/journal.pone.0276726 · Cited by 5
“MIL has been adopted successfully for microscopy-specific tasks such as classifying and segmenting cells [21] with recent reviews [22], [23] detailing the different approaches.”
Method
Multiple Instance Learning for Offensive Language Detection
Liu, J., Kong, D., Huang, L., Mao, D., & Xue, H. (2022) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2022.findings-emnlp.546 · Cited by 5
“Previous studies(Carbonneau et al., 2018) have shown that tasks with lower WR are more challenging, so MINOR could be easier for a model even if it’s automatically labeled.”
Background
SQ-Swin: Siamese Quadratic Swin Transformer for Lettuce Browning Prediction
Wang, D., Zhang, B., Xu, Y., Luo, Y., & Yu, H. (2023) · IEEE Access · doi:10.1109/access.2023.3332488 · Cited by 9
Regularized Instance Embedding for Deep Multi-Instance Learning
Lin, Y., & Zhang, H. (2019) · Applied Sciences · doi:10.3390/app10010064 · Cited by 5
“As pointed out by Carbonneau et al. [7], in positive bags, some false-negative and false-positive instances barely affect bag classiﬁcation.”
BackgroundMethodInfluential
Decompose the Sounds and Pixels, Recompose the Events
Rao, V. R., Khalil, M. I., Li, H., Dai, P., & Lu, J. (2022) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v36i2.20111 · Cited by 5
PROMIPL: A Probabilistic Generative Model for Multi-Instance Partial-Label Learning
Yang, Y.-F., Tang, W., & Zhang, M.-L. (2024) · Proceedings · doi:10.1109/icdm59182.2024.00063 · Cited by 4
FAWL: Weakly-Supervised Video Corpus Moment Retrieval with Frame-Wise Auxiliary Alignment and Weighted Contrastive Learning
Pan, Y. F., Zhang, Y., & Zhao, X. (2025) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp49660.2025.10887823 · Cited by 4
“It proposes a baseline method, JSG, which is based on Contrastive Learning [6](CL) and Multiple Instance Learning [7, 8](MIL) frameworks.”
Multiple instance learning-based prediction of programmed death-ligand 1 (PD-L1) expression from hematoxylin and eosin (H&E)-stained histopathological images in breast cancer
Da, Z., Yang, H., Zhaxi, B., Sun, K., Bai, G., Wang, C., Wang, F., Pan, W., & Du, R. (2025) · PeerJ · doi:10.7717/peerj.19201 · Cited by 4
“These methods train models with minimal labeled data, effectively addressing data bottlenecks in medical image analysis, and have achieved some progress in pathology image analysis ( Carbonneau et al., 2018 ).”
Fuzzy Optimized Attention Network with Multi-Instance Deep Learning (FOAN-MIDL) for Alzheimer’s Disease Diagnosis with Structural Magnetic Resonance Imaging (sMRI)
Alhassan, A. M., & Altmami, N. I. (2025) · Diagnostics · doi:10.3390/diagnostics15121516 · Cited by 4
“The MIL approach to medical image analysis is employed since, typically, medical image datasets are small, with weak labeling but a reasonably high resolution [13].”
Automatic labels are as effective as manual labels in digital pathology images classification with deep learning
Marini, N., Marchesin, S., Ferrís, L. B., Püttmann, S., Wodziński, M., Fratti, R., Podareanu, D., Caputo, A., Boytcheva, S., Vatrano, S., et al. (2025) · Journal of Pathology Informatics · doi:10.1016/j.jpi.2025.100462 · Cited by 7
“Currently, most weakly supervised algorithms in computational pathology are based on the multiple instance learning (MIL) approach 6 that models the whole image as a bag of instances, where only global annotations are available.”
Multiple-instance Learning from Triplet Comparison Bags
Shu, S., Wang, D.-B., Yuan, S., Wei, H., Jiang, J., Feng, L., & Zhang, M.-L. (2024) · ACM Transactions on Knowledge Discovery from Data · doi:10.1145/3638776 · Cited by 4
“However, the instance-level methods are usually worse than the bag-level methods as shown in previous work [9].”
BackgroundMethod
Integrative habitat analysis and multi-instance deep learning for predictive model of PD-1/PD-L1 immunotherapy efficacy in NSCLC patients: a dual-center retrospective study
Huang, X., Huang, X., Xie, Y., Wang, K., Bai, H., Ning, R., Zhu, X., Huang, D., & Jin, G. (2025) · BMC Medical Imaging · doi:10.1186/s12880-025-01828-5 · Cited by 4
“Multiple instance learning (MIL) is a weakly supervised learning that not only eliminates the need for instance-level labels, thus significantly reducing labeling costs, but also organically fuses instance features to improve the predictive performance of the model [25].”
Functional Site Discovery From Incomplete Training Data: A Case Study With Nucleic Acid–Binding Proteins
Wang, W., Langlois, R. E., Langlois, M., Genchev, G. Z., Wang, X., & Lu, H. J. (2019) · Frontiers in Genetics · doi:10.3389/fgene.2019.00729 · Cited by 4
“…have been developed to deal with incomplete knowledge of the training data such as: semi-supervised learning (Chapelle et al., 2010), active learning (Reker and Schneider, 2015), positive and unlabeled learning (Bhardwaj et al., 2010), and multiple-instance learning (MIL) (Carbonneau et al., 2018).”
Background
Audio-Visual Weakly Supervised Approach for Apathy Detection in the Elderly
Sharma, G., Joshi, J., Zeghari, R., & Guerchouche, R. (2020) · International Joint Conference on Neural Networks (IJCNN) · doi:10.1109/ijcnn48605.2020.9206829 · Cited by 4
“In MIL, the training data is divided into multiple sets and one label is used for the complete set without having the labels for each data point [30].”
Method
In Defense of LSTMs for Addressing Multiple Instance Learning Problems
Wang, K., Jose Oramas, M., & Tuytelaars, T. (2021) · Lecture notes in computer science · doi:10.1007/978-3-030-69544-6_27 · Cited by 5

Authors: Carbonneau, M.-A., Cheplygina, V., Granger, E., & Gagnon, G.
Affiliations: École de Technologie Supérieure (CA)
Erasmus MC (NL)
Corresponding author: Carbonneau, Marc-André
ORCID: Carbonneau, Marc-André
Cheplygina, Veronika
Gagnon, Ghyslain
Venue: Pattern Recognition
Publisher: Elsevier BV
Published: 2018-05
Volume: 77
Pages: 329-353
ISSN: 0031-3203 (ISSN-L)
DOI: 10.1016/j.patcog.2017.10.009
OpenAlex: W2560886373
Cited by: 740
References: 223
Citations by year: 2026: 75 · 2025: 118 · 2024: 91 · 2023: 124 · 2022: 98 · 2021: 92 · 2020: 84 · 2019: 41 · 2018: 14
FWCI: 22.10
Type: journal article
Language: en
Open access: green
arxiv.org
Licence: www.elsevier.com/tdm/userlicense/1.0/ (tdm)
Locations: Pattern Recognition (publishedVersion)
arXiv (Cornell University) (submittedVersion)
TU/e Research Portal (submittedVersion)
EUR Research Repository (Erasmus University Rotterdam) (submittedVersion)
Espace ÉTS (ETS) (submittedVersion)
TU/e Research Portal (submittedVersion, other-oa)
RePub (Erasmus University Rotterdam) (submittedVersion, other-oa)
arXiv (Cornell University)
Topics: Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Algorithms
Physical Sciences › Computer Science › Artificial Intelligence › Machine Learning and Data Classification
Physical Sciences › Computer Science › Artificial Intelligence › Domain Adaptation and Few-Shot Learning
Keywords: multiple instance learning, weakly supervised learning, problem characteristics, data distribution, label ambiguity, computer vision, document classification

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

Multiple instance learning: A survey of problem characteristics and applications

Overview

Integrity signals

Verification

Claims the paper makes

Claims other papers make about this one

Self-check

Citations

Cites

Cited by

Related

Metadata

Record

About these checks