Abstract
While AI is expanding to many systems and services from search engines to online retail, a revolution is needed, to produce rapid, reliable “AI everywhere” applications by “continuous, cross-domain learning”. We introduce Synthesizable Artificial Intelligence, and discuss its uniqueness by its five advanced “abilities”; (1) continuous learning after training by “connecting the dots”; (2) measuring quality of success; (3) correcting concept drift; (4) “self-correcting” for new paradigms; and (5) retroactively applying new learning for development of “long-term self-learning”. SAI can retroactively apply new concepts to old examples, “self-learning” in a new way by considering recent experiences similar to the human experience. We demonstrate its current and future applications in transferring seamlessly from one domain to another, and show its use in commercial applications, including engine sound analysis, providing real-time indications of potential engine failure.
The authors acknowledge the sponsorship of NASA Ames Research Center, US Department of Agriculture, National Science Foundation, US Department of Defense and the US Army Research Lab in their research.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
While deep learning techniques have eliminated the need for automatically extracting features, they have been shown not to work well, for example, for texture datasets where the inherent dimensionality of the data is high [2].
- 2.
Some individual pieces of the puzzle are already developed in subfields of AI like active learning and transfer learning.
References
Iyengar, S., Mukhopadhyay, S., Steinmuller, C., Li, X.: Preventing future oil spills with software-based event detection. IEEE Comput. 43(8), 95–97 (2010). IEEE Computer Society, 0018–9162/10
Karki, M., DiBiano, R., Basu, S., Mukhopadhyay, S.: Core sampling framework for pixel classification. In: Proceedings of the International Conference on Artificial Neural Networks (2017)
Basu, S., Karki, M., Mukhopadhyay, S., Ganguly, S., Nemani, R., DiBiano, R., Gayaka, S.: A theoretical analysis of Deep Neural Networks for texture classification. IJCNN 2016, 992–999 (2016)
DiBiano, R., Mukhopadhyay, S.: Automated diagnostics for manufacturing machinery based on well regularized deep neural networks, integration. VLSI J. 58, 303–310 (2017)
Basu, S., Ganguly, S., Nemani, R., Mukhopadhyay, S., Zhang, G., Milesi, C., et al.: A semi automated probabilistic framework for tree cover delineation from 1-M NAIP imagery using a high performance computing architecture. IEEE Trans. Geosci. Remote Sens. 53(10), 5690–5708 (2015)
Basu, S., Ganguly, S., Mukhopadhyay, S., DiBiano, R., Karki, M., Nemani, R.: DeepSat—a learning framework for satellite imagery. In: Proceedings of the ACM SIGSPATIAL 2015 (2015)
Sidhanta, S., Golab, W., Mukhopadhyay, S., Basu, S.: Adaptable SLA-aware consistency tuning for quorum-replicated data stores. IEEE Trans. Big Data 3, 248–261 (2017)
Sidhanta, S., Mukhopadyay, S.: Infra: SLO aware elastic auto scaling in the cloud for cost reduction. In: IEEE BigData Congress, pp. 141–148 (2016)
Alvin, C., Gulwani, S., Majumdar, R., Mukhopadhyay, S.: Synthesis of geometry proof problems. In: Proceedings of AAAI, pp. 245–252 (2014)
Alvin, C., Gulwani, S., Majumdar, R., Mukhopadhyay, S.: Synthesis of solutions for shaded area geometry problems. In: Proceedings of FLAIRS (2017)
Naderi, M., Alvin, C., Ding, Y., Mukhopadhyay, S., Brylinski, M.: A graph-based approach to construct target focused libraries for virtual screening. J. Chemoinform. 8, 14 (2016)
Alvin, C., Peterson, B., Mukhopadhyay, S.: StaticGen: static generation of UML sequence diagrams. In: Proceedings of the International Conference on the Foundational Aspects of Software Engineering (2017)
Mukhopadhyay, S., Iyengar, S.S.: System and architecture for robust management of resources in a wide-area network. US Patent Number 9,240,955 issued January 2016
Alvin, C., Gulwani, S., Majumdar, R., Mukhopadhyay, S.: Synthesis of problems for shaded area geometry reasoning. In: Proceedings of AIED (2017)
Basu, S., Karki, M., Ganguly, S., DiBiano, R., Mukhopadhyay, S., Gayaka, S., Kannan, R., Nemani, R.: Learning sparse feature representations using probabilistic quadtrees and deep belief nets. Neural Process. Lett. 1–13 (2016). https://doi.org/10.1007/s11063-016-9556-4
Liu, T., Naderi, M., Alvin, C., Mukhopadhyay, S., Brylinski, M.: Break down in order to build up: decomposing small molecules for fragment-based drug design with eMolFrag. J. Chem. Inf. Model. 57, 627–631 (2017)
Boyda, E., Basu, S., Ganguly, S., Michaelis, A., Mukhopadhyay, S., Nemani, R.: Deploying a quantum annealing processor to detect tree cover in aerial imagery of California. PLoS ONE (2017)
Ganguly, S., Basu, S., Nemani, R., Mukhopadhyay, S., Michaelis, A., Votava, P., Milesi, C., Kumar, U.: Deep learning for very high resolution imagery classification. In: Srivastava, A., Nemani, R., Steinhaeuser, K. (eds.) Large-Scale Machine Learning in the Earth Sciences. CRC Press, Boca Raton (2017)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2019 Springer Nature Switzerland AG
About this paper
Cite this paper
Mukhopadhyay, S., Iyengar, S.S., Madni, A.M., Di Biano, R. (2019). The Next Generation of Artificial Intelligence: Synthesizable AI. In: Arai, K., Bhatia, R., Kapoor, S. (eds) Proceedings of the Future Technologies Conference (FTC) 2018. FTC 2018. Advances in Intelligent Systems and Computing, vol 880. Springer, Cham. https://doi.org/10.1007/978-3-030-02686-8_50
Download citation
DOI: https://doi.org/10.1007/978-3-030-02686-8_50
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-02685-1
Online ISBN: 978-3-030-02686-8
eBook Packages: Intelligent Technologies and RoboticsIntelligent Technologies and Robotics (R0)