Home / Current Issue / Paper 1719862
Neuro-Symbolic AI: Current Progress and Future Directions
Subject area: Science,Engineering and Technology · Area of research: Neuro-Symbolic Artificial Intelligence
Abstract
Deep learning has delivered remarkable perceptual accuracy across vision, speech, and language tasks, yet it continues to struggle with systematic generalization, interpretability, and the incorporation of prior knowledge. Symbolic artificial intelligence, in contrast, offers transparent reasoning and compositional structure but scales poorly to noisy, high-dimensional data. Neuro-symbolic AI seeks to combine the statistical learning capacity of neural networks with the structured, verifiable reasoning of symbolic systems. This paper surveys the field's foundations, presents a taxonomy of integration strategies, and reviews representative architectures including Logic Tensor Networks, DeepProbLog, Neural Theorem Provers, and the Neuro-Symbolic Concept Learner. We examine applications in visual question answering, natural language understanding, robotics, and scientific reasoning, summarize commonly used benchmarks, and discuss open challenges related to scalability, differentiable reasoning, and knowledge representation. We conclude by outlining research directions that connect neuro-symbolic methods with large language models and program synthesis, arguing that hybrid architectures are likely to play a growing role in building AI systems that are both capable and explainable.
Keywords
Neuro-Symbolic AI, Symbolic Reasoning, Deep Learning, Knowledge Representation, Explainable AI, Logic Tensor Networks, Probabilistic Logic Programming
References
[1] A. d'Avila Garcez, T. R. Besold, L. De Raedt, P. Foldiak, P. Hitzler, T. Icard, K.-U. Kuhnberger, L. C. Lamb, R. Miikkulainen, and D. L. Silver, "Neural-symbolic learning and reasoning: Contributions and challenges," in Proc. AAAI Spring Symp. Series, 2015.
[2] T. R. Besold, A. d'Avila Garcez, S. Bader, H. Bowman, P. Domingos, P. Hitzler, K.-U. Kuhnberger, L. C. Lamb, D. Lowd, P. M. V. Lima, L. de Penning, G. Pinkas, H. Poon, and G. Zaverucha, "Neural-symbolic learning and reasoning: A survey and interpretation," arXiv preprint, 2017.
[3] L. Serafini and A. d'Avila Garcez, "Logic Tensor Networks: Deep learning and logical reasoning from data and knowledge," in Proc. NeSy Workshop, 2016.
[4] S. Badreddine, A. d'Avila Garcez, L. Serafini, and M. Spranger, "Logic Tensor Networks," Artificial Intelligence, vol. 303, 2022.
[5] R. Manhaeve, S. Dumancic, A. Kimmig, T. Demeester, and L. De Raedt, "DeepProbLog: Neural probabilistic logic programming," in Proc. NeurIPS, 2018.
[6] T. Rocktaschel and S. Riedel, "End-to-end differentiable proving," in Proc. NeurIPS, 2017.
[7] J. Mao, C. Gan, P. Kohli, J. B. Tenenbaum, and J. Wu, "The Neuro-Symbolic Concept Learner: Interpreting scenes, words, and sentences from natural supervision," in Proc. ICLR, 2019.
[8] K. Yi, J. Wu, C. Gan, A. Torralba, P. Kohli, and J. B. Tenenbaum, "Neural-symbolic VQA: Disentangling reasoning from vision and language understanding," in Proc. NeurIPS, 2018.
[9] Z. Li, C. Chen, X. Guo, and S. Zhang, "Scallop: A language for neurosymbolic programming," in Proc. PLDI, 2023.
[10] Z. Yang, A. Ishay, and J. Lee, "NeurASP: Embracing neural networks into answer set programming," in Proc. IJCAI, 2020.
[11] L. De Raedt, A. Kimmig, and H. Toivonen, "ProbLog: A probabilistic Prolog and its application in link discovery," in Proc. IJCAI, 2007.
[12] G. Marcus, "The next decade in AI: Four steps towards robust artificial intelligence," arXiv preprint, 2020.
[13] B. M. Lake, T. D. Ullman, J. B. Tenenbaum, and S. J. Gershman, "Building machines that learn and think like people," Behavioral and Brain Sciences, vol. 40, 2017.
[14] A. d'Avila Garcez and L. C. Lamb, "Neurosymbolic AI: The 3rd wave," arXiv preprint, 2020.
[15] S. Chaudhuri, K. Ellis, O. Polozov, R. Singh, A. Solar-Lezama, and Y. Yue, "Neurosymbolic programming," Foundations and Trends in Programming Languages, vol. 7, 2021.
[16] R. Riegel, A. Gray, F. Luus, N. Khan, N. Makondo, I. Y. Akhalwaya, H. Qian, R. Fagin, F. Barahona, U. Sharma, S. Ikbal, H. Karanam, S. Neelam, A. Likhyani, and S. Srivastava, "Logical neural networks," arXiv preprint, 2020.
[17] P. Hitzler and M. K. Sarker, Eds., Neuro-Symbolic Artificial Intelligence: The State of the Art, IOS Press, 2021.
[18] D. Kahneman, Thinking, Fast and Slow. New York, NY, USA: Farrar, Straus and Giroux, 2011.
[19] J. Yang, L. Li, and X. Chen, "A survey on neuro-symbolic learning systems," Neural Networks, vol. 166, 2023.
[20] S. Dash, K. Chitlangia, A. Ahuja, and A. Srinivasan, "A review of some techniques for inclusion of domain-knowledge into deep neural networks," Scientific Reports, vol. 12, 2022.
[21] M. Yin, P. Xu, and H. Wang, "Neuro-symbolic reasoning for compositional generalization: A survey," ACM Computing Surveys, 2024.
[22] S. Amizadeh, H. Palangi, A. Polozov, Y. Huang, and K. Koishida, "Neuro-symbolic visual reasoning: Disentangling visual and reasoning," in Proc. ICML, 2020.
[23] K. Ellis, D. Ritchie, A. Solar-Lezama, and J. Tenenbaum, "Learning to infer graphics programs from hand-drawn images," in Proc. NeurIPS, 2018.
[24] R. Evans and E. Grefenstette, "Learning explanatory rules from noisy data," Journal of Artificial Intelligence Research, vol. 61, 2018.
[25] H. Dong, J. Mao, T. Lin, C. Wang, L. Li, and D. Zhou, "Neural logic machines," in Proc. ICLR, 2019.
[26] W. W. Cohen, F. Yang, and K. Mazaitis, "TensorLog: Deep learning meets probabilistic databases," Journal of AI Research, 2020.
[27] L. C. Lamb, A. Garcez, M. Gori, M. Prates, P. Avelar, and M. Vardi, "Graph neural networks meet neural-symbolic computing: A survey and perspective," in Proc. IJCAI, 2020.
[28] N. Muggleton and L. De Raedt, "Inductive logic programming: Theory and methods," Journal of Logic Programming, vol. 19, 1994.
[29] F. Yang, Z. Yang, and W. W. Cohen, "Differentiable learning of logical rules for knowledge base reasoning," in Proc. NeurIPS, 2017.
[30] T. Demeester, T. Rocktaschel, and S. Riedel, "Lifted rule injection for relation embeddings," in Proc. EMNLP, 2016.
[31] H. A. Kautz, "The third AI summer: AAAI Robert S. Engelmore Memorial Lecture," AI Magazine, vol. 43, 2022.
[32] M. Mitchell, "Abstraction and analogy-making in artificial intelligence," Annals of the New York Academy of Sciences, vol. 1505, 2021.
[33] P. Hohenecker and T. Lukasiewicz, "Ontology reasoning with deep neural networks," Journal of Artificial Intelligence Research, vol. 68, 2020.
[34] V. Belle, "Symbolic logic meets machine learning: A brief survey in infinite domains," in Proc. SUM, 2020.
[35] A. Sinha, P. Rai, and S. Mausam, "Neuro-symbolic approaches for text-based games and interactive fiction," in Proc. EMNLP, 2022.
[36] J. Andreas, M. Rohrbach, T. Darrell, and D. Klein, "Neural module networks," in Proc. CVPR, 2016.
[37] R. Riegel and A. Gray, "Neuro-symbolic AI for enterprise reasoning applications," IBM Journal of Research and Development, 2021.
[38] A. Yang, D. Ritter, and V. Belle, "Combining rule-based and statistical reasoning in knowledge graphs: A survey," Journal of Web Semantics, vol. 71, 2023.
[39] L. Lamb, R. Baldo, and A. Garcez, "Reasoning about generalization via neural-symbolic integration," in Proc. IJCAI Survey Track, 2021.
[40] T. Dinu, A. Bogdan, and C. Popescu, "Neuro-symbolic reasoning for robot task planning: A review," Robotics and Autonomous Systems, vol. 160, 2023.
[41] M. Nye, A. Solar-Lezama, J. Tenenbaum, and B. Lake, "Learning compositional rules via neural program synthesis," in Proc. NeurIPS, 2020.
[42] S. Ren, K. Cho, and S. Wiseman, "Neuro-symbolic language grounding for instruction following," in Proc. ACL, 2022.
[43] P. Minervini, S. Riedel, P. Stenetorp, E. Grefenstette, and T. Rocktaschel, "Learning reasoning strategies in end-to-end differentiable proving," in Proc. ICML, 2020.
[44] V. Gupta, M. Chaudhary, and R. Sharma, "Neuro-symbolic AI in cybersecurity: Opportunities and challenges," Computers and Security, vol. 128, 2023.
How to cite this paper
@article{1719862,
author = {Shyam Sundar, Kanika Rana, Jatin, Vivek, Divya Singla},
title = {Neuro-Symbolic AI: Current Progress and Future Directions},
journal = {Iconic Research And Engineering Journals},
year = {2024},
volume = {8},
number = {2},
pages = {1416-1427},
issn = {2456-8880},
url = {https://www.irejournals.com/formatedpaper/1719862.pdf},
abstract = {Deep learning has delivered remarkable perceptual accuracy across vision, speech, and language tasks, yet it continues to struggle with systematic generalization, interpretability, and the incorporation of prior knowledge. Symbolic artificial intelligence, in contrast, offers transparent reasoning and compositional structure but scales poorly to noisy, high-dimensional data. Neuro-symbolic AI seeks to combine the statistical learning capacity of neural networks with the structured, verifiable reasoning of symbolic systems. This paper surveys the field's foundations, presents a taxonomy of integration strategies, and reviews representative architectures including Logic Tensor Networks, DeepProbLog, Neural Theorem Provers, and the Neuro-Symbolic Concept Learner. We examine applications in visual question answering, natural language understanding, robotics, and scientific reasoning, summarize commonly used benchmarks, and discuss open challenges related to scalability, differentiable reasoning, and knowledge representation. We conclude by outlining research directions that connect neuro-symbolic methods with large language models and program synthesis, arguing that hybrid architectures are likely to play a growing role in building AI systems that are both capable and explainable.},
keywords = {Neuro-Symbolic AI, Symbolic Reasoning, Deep Learning, Knowledge Representation, Explainable AI, Logic Tensor Networks, Probabilistic Logic Programming},
month = {August},
}