A Review on Explainability in Multimodal Deep Neural Nets
Open accessThe DOI matches its Crossref record; 268 of the 268 references it cites were found in a registry or an index.
Abstract
Artificial Intelligence techniques powered by deep neural nets have achieved much success in several application domains, most significantly and notably in the Computer Vision applications and Natural Language Processing tasks. Surpassing human-level performance propelled the research in the applications where different modalities amongst language, vision, sensory, text play an important role in accurate predictions and identification. Several multimodal fusion methods employing deep learning models are proposed in the literature. Despite their outstanding performance, the complex, opaque and black-box nature of the deep neural nets limits their social acceptance and usability. This has given rise to the quest for model interpretability and explainability, more so in the complex tasks involving multimodal AI methods. This paper extensively reviews the present literature to present a comprehensive survey and commentary on the explainability in multimodal deep neural nets, especially for the vision and language tasks. Several topics on multimodal AI and its applications for generic domains have been covered in this paper, including the significance, datasets, fundamental building blocks of the methods and techniques, challenges, applications, and future trends in this domain.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
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?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“The black box and opaque nature of deep neural networks precludes their usefulness despite excellent performance of multimodal methods [288].”
From Survey of explainable artificial intelligence techniques for biomedical imaging with deep neural networks (2023) · cites this paper as [288]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“AI systems can profit significantly from assimilating multimodal data into prediction and classification models to imitate integrative human clinical decision-making.”
From Orchestrating explainable artificial intelligence for multimodal and longitudinal data in medical imaging (2024) · cites this paper as [17]
Experimental — accuracy not yet measuredThe source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“This can boost their robustness and accuracy, enable the discovery of new biomarkers and therapeutic targets6,14, as well as improve model performance15–17.”
From Orchestrating explainable artificial intelligence for multimodal and longitudinal data in medical imaging (2024) · cites this paper as [17]
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, L., & Polosukhin, I. (2017) · arXiv · doi:10.48550/arxiv.1706.03762 · Cited by 120,404 · cited 1×
“[165] Attention Mechanism Focuses and attends the regions that are important for making”
BackgroundHochreiter, S., & Schmidhuber, J. (1997) · Neural Computation · doi:10.1162/neco.1997.9.8.1735 · Cited by 102,318 · cited 1×
“BERT Transformer (B2T2) [98], UnicodeText Unidirectional long short term memory(LSTM)[104], Bidirectional LSTM (BiLSTM), Unidirectional Gated Recurrent Units (GRU)[105],bidirectional GRU (BiGRU) VL [99], VLBERT[100], UNITER [101], ViLBert [102], LXMERT [103]”
BackgroundLeCun, Y., Bengio, Y., & Hinton, G. (2015) · Nature · doi:10.1038/nature14539 · Cited by 84,126
Exploiting Generative AI to Scale up Intelligent Tutoring Systems
Jakubův, J., Chvalovský, K., Goertzel, Z., Kaliszyk, C., Olšák, M., Piotrowski, B., Schulz, S., Suda, M., & Urban, J. (2023) · DROPS (Schloss Dagstuhl – Leibniz Center for Informatics) · doi:10.4230/lipics.itp.2023.19 · Cited by 76,658
Rich Feature Hierarchies for Accurate Object Detection and Semantic Segmentation
Girshick, R., Donahue, J., Darrell, T., & Malik, J. (2014) · 2014 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2014.81 · Cited by 32,684
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 3×
“Individual conditional expectation (ICE) plot [145] visualize the dependence of features on the final prediction for unique data points, providing more insights than the PDP approach of averaging on all data points.”
BackgroundA Unified Approach to Interpreting Model Predictions
Lundberg, S., & Lee, S.-I. (2017) · arXiv · doi:10.48550/arxiv.1705.07874 · Cited by 23,707 · cited 1×
“Shapley values [146] are used when each data point’s contribution is variable in the final prediction.”
MethodBLEU: a method for automatic evaluation of machine translation
Papineni, K., Roukos, S., Ward, T., & Zhu, W.-J. (2001) · Proceedings of the 40th Annual Meeting on Association for Computational Linguistics - ACL '02 · doi:10.3115/1073083.1073135 · Cited by 21,344 · cited 2×
“Automated evaluation metrics such as BLEU-4 [215] is used for automatic evaluation of machine translation that is”
Method"Why Should I Trust You?": Explaining the Predictions of Any Classifier
Ribeiro, M. T., Singh, S., & Guestrin, C. (2016) · Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining · doi:10.1145/2939672.2939778 · Cited by 17,555
Visualizing and understanding convolutional networks
Zeiler, M. D., & Fergus, R. (2014) · Lecture Notes in Computer Science · doi:10.1007/978-3-319-10590-1_53 · Cited by 15,839 · cited 1×
“Deep learning-based explanatory systems are broadly classified into justification [17] or introspection-based systems [155], [165].”
Background
Research works printed but not linked: 13
- [179] J. Strout, Y. Zhang, and R. J. Mooney, 'Do Human Rationales Improve Machine Explanations?,' arXiv Prepr. arXiv1905.13714 , 2019. -
- [184] S. Vidya Sagar Appaji, P. V. Lakshmi, and P. Srinivasa Rao, 'Maximizing joint probability in visual question answering models,' Int. J. Adv. Sci. Technol. , vol. 29, no. 3, pp. 39143923, 2020. -
- [84] T. Showcases, 'Unbox the Black-box for the Medical Explainable AI via Multi-modal and Multi-centre Data Fusion :' -
- [206] J. Lin, U. Jain, and A. G. Schwing, 'TAB-VCR: Tags and attributes based VCR baselines,' arXiv , no. NeurIPS, 2019. -
- [100] W. Su et al. , 'VL-BERT: Pre-training of generic visuallinguistic representations,' arXiv , pp. 1-16, 2019. -
- [199] D. Braines and D. Harborne, 'Multimodal Explanations for AIbased Multisensor Fusion,' pp. 1-13. -
- tharwat, 2020, Appl Comput Inform
- [219] M. ul Hassan, P. Mulhem, D. Pellerin, and G. Quénot, 'Explaining Visual Classification using Attributes,' in 2019 International Conference on Content-Based Multimedia Indexing (CBMI) , 2019, pp. 1-6. -
- [230] U. E. Ai, 'Can we do better explanations ? A proposal of.' -
- [232] P. Welinder, S. Branson, T. Mita, C. Wah, and F. Schroff, 'Caltech-ucsd Birds 200,' Caltech-UCSD Tech. Rep. , vol. 200, pp. 1-15, 2010, [Online]. Available: http://www.flickr.com/. -
- [236] V. Sharma, A. Kalra, S. Chaudhary, L. Patel, and L. P. Morency, 'Attend and Attack : Attention Guided Adversarial Attacks on Visual Question Answering Models,' 32nd Conf. Neural Inf. Process. Syst. , no. NeurIPS, pp. 1-6, 2018. -
- [142] O. Biran and C. Cotton, 'Explanation and justification in 18. machine learning: A survey,' in IJCAI-17 workshop on explainable AI (XAI) , 2017, vol. 8, no. 1, pp. 8-13. -
- [112] D. Kiela, E. Grave, A. Joulin, and T. Mikolov, 'Efficient largescale multi-modal classification,' 32nd AAAI Conf. Artif. Intell. AAAI 2018 , pp. 5198-5204, 2018. -
Cited by
Items 201–222 of 222
Cross-Modal Rationale Transfer for Explainable Humanitarian Classification on Social Media
Nguyen, T. H., Rudra, K., & Nejdl, W. (2026) · International World Wide Web Conference (WWW) · doi:10.1145/3774904.3792991
“LLM-as-a-judge has shown its on-par performance with human evaluators and has been widely used for evaluation tasks [71].”
Deepika, N., & Renuka, D. K. (2026) · International Conference on Intelligent Systems for Communication, IoT and Security (ICISCoIS) · doi:10.1109/iciscois62701.2026.11447909
“Interpreting how multimodal systems arrive at a conclusion is another important challenge that is essential to the reliability and explainability of AI [12][13].”
A Self-Supervised Attribution Method for Explaining Neural Networks
Xia, T. (n.d.) · AMS Degree Thesis (University of Bologna)
Enrichissement linguistique et cognitif de modèles de langue contextualisés pour le domaine médical
Blanc, C. (2023) · theses.fr (ABES) · doi:10.70675/0b45b716z1b6ez4362z8081z4731e0fb159d
Ghaoui, M. A. (2023) · theses.fr (ABES) · doi:10.70675/ba3b720bz66cfz4cedzbc73zf478b1c0dee1
Classification de données médicales textuelles et structurées par apprentissage profond
Bailly, A. (2023) · theses.fr (ABES) · doi:10.70675/897fe067zf4dcz488az8c42za207b15f7daa
Fault Diagnosis of Track Circuits Based on Fuzzy Broad Learning System
Li, L., Huang, Z., Yang, Y., Chen, M., Wang, J., & Hao, J. (2025) · China Automation Congress (CAC) · doi:10.1109/cac67268.2025.11487749
Neural Networks in Electrical Engineering Applications: A Review
Khan, F. (2026) · Global Trends in Science and Technology · doi:10.70445/gtst.2.2.2026.219-239 · Cited by 3
Predicting liquefaction potential from seismic events
Geem, Z. W., Bekdaş, G., Nigdeli, S. M., Aydın, Y., Işıkdağ, Ü., & Kim, T.-H. (2026) · Elsevier eBooks · doi:10.1016/b978-0-443-41481-7.00009-2
Hybrid Digital Forensic Framework for Multimodal Deepfake DetectionPreprint
Qureshi, S. M., Saeed, A., Ahmad, F., Ghamdi, M. A. A., & Alsolbi, I. N. (2026) · Research Square · doi:10.21203/rs.3.rs-9312223/v1
Quantitative relationship mining of steel CPSP through multimodal data fusion
Zhang, C., Cao, G., Wu, S., Yan, X., Xiong, W., Wang, H., Liu, N., & Liu, Z. (2026) · Journal of Materials Science · doi:10.1007/s10853-026-13055-6
“By leveraging deep neural networks to construct cross-modal feature mappings and embedding physical knowledge, these methods enhance model interpretability and generalization capability [25].”
Models and Modeling: The Essence of AI Development
Schmid, T., Hildesheim, W., & Holoyad, T. (2026) · Springer eBooks · doi:10.1007/978-3-032-00533-5_2
Prescriptive modality selection for classification and regression
Carrizosa, E., Castro, J. C., & Guerrero, V. (2026) · European Journal of Operational Research · doi:10.1016/j.ejor.2026.07.009 · Cited by 1
Emerging trends and technologies in explainable AI (XAI) for clinical practice
Nazreen, Thayanithi, C. A., Arjun, E., Singh, P., & Hetu, V. (2026) · Elsevier eBooks · doi:10.1016/b978-0-443-45374-8.00012-5
A Self-correcting Neuro-symbolic AI Reasoning Framework
Wooding, B., Brennan, K., Tumlin, A. M., Zhang, H., & Johnson, T. (2026) · Lecture notes in computer science · doi:10.1007/978-3-032-32357-6_9 · Cited by 1
Multimodal Document Classification Across Domains and Methods: A Systematic Review
Aljuhani, H. L., Dahab, M. Y., & Alsenani, Y. S. (2026) · Engineering Technology & Applied Science Research · doi:10.48084/etasr.19175
Sánchez, J. G. H., Cruz-Olguín, J. M., Santos-Romero, S., Árcega, A. F., Morales, V. L., & Ojeda-Misses, M. A. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.7315921
Abbour, F. Z., Belarache, O., Ardchir, S., Ounacer, S., & Azzouazi, M. (2026) · International Conference on Circuit, Systems and Communication (ICCSC) · doi:10.1109/iccsc71566.2026.11650302
“[24], [25], [26] contributed to this line of research, creating multimodal models utilizing psychometrics, texts, demographics, and exhibiting improved predictive power with explainability problems.”
An ECG-EHR In-Hospital Mortality Risk Prediction Method Based on Multimodal Fusion
Zhang, J., Wei, Z., Liu, S., 林国秀, Sun, L., & He, F. (2026) · International Conference on Health Big Data · doi:10.1145/3833450.3833471
“Integrating ECG, EHR, and imaging data can significantly enhance the accuracy and robustness of decision systems [22-23].”
Explainable Artificial Intelligence (XAI): Shedding Light on AI's Black Box
Gandhi, K. (2024) · International Journal of Intelligent Systems and Applications in Engineering
Research on multimodal medical data fusion and application based on large-scale models
Zhao, N., Li, Y.-Q., & Liu, H. (2026) · Journal of Chinese Integrative Medicine · doi:10.1016/j.joim.2026.11.002
Metadata
- Authors
- Joshi, G., Walambe, R., & Kotecha, K.
- Affiliations
- Symbiosis International University (IN)
- Venue
- IEEE Access
- Published
- 2021
- Volume
- 9
- Pages
- 59800-59821
- ISSN
- 2169-3536 (ISSN-L)
- OpenAlex
- W3148438775
- Semantic Scholar
- 6bda005a727969356a012e177cd600e7fdf80c96
- Cited by
- 204
- References
- 261
- Citations by year
- 2026: 26 · 2025: 49 · 2024: 45 · 2023: 50 · 2022: 25 · 2021: 9
- FWCI
- 11.40
- Type
- journal article
- Language
- en
- Open access
- gold
- doi.org
- Locations
- IEEE Access (publishedVersion, cc-by)
- arXiv (Cornell University) (submittedVersion)
- DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
- arXiv (Cornell University) (cc-by)
- Funders
- Ministry of Human Resource Development (MHRD) Scheme for Promotion of Academic and Research Collaboration (SPARC) of the Government of India (P104)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Multimodal Machine Learning Applications
- Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Advanced Neural Network Applications
- Keywords
- explainability, multimodal deep neural networks, multimodal fusion, model interpretability, vision-language tasks, black-box models
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.