From trivial representations to learning concepts in AI by exploiting unique data

Project Details

Description

The prospect of an AI-based revolution and its socio-economic benefits is tantalising. We want to believe that AI learns higher level concepts, but this is far from reality. Particularly in unstructured data such as images, AI extracts rather trivial notions even when provided with millions of examples. The solution of “the more plentiful and diverse the data the merrier” will not address this. Beyond the privacy and cost implications of amassing data in key applications (e.g. healthcare), some events (e.g. disease) can be rare or truly unique. However, the data inefficiency of modern AI also manifests in poorly leveraging the goldmine of information present in unique and rare data. The key research question we aim to answer is Why does AI struggle with concepts and what is the role of unique data? We propose to move away from trivial learning to a formal representation and learning of concepts that overhauls machine learning precisely by leveraging unique data. This paradigm shift will create unprecedented rewards compared to the current status quo
StatusActive
Effective start/end date1/02/2331/01/25

Fingerprint

Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.
  • Neural networks memorise personal information from one sample

    Hartley, J., Sanchez, P., Haider, F. & Tsaftaris, S. A., Dec 2023, In: Scientific Reports. 13, 1, p. 21366 21366.

    Research output: Contribution to journalArticlepeer-review

    Open Access
  • Group Distributionally Robust Knowledge Distillation

    Vilouras, K., Liu, X., Sanchez, P., O’Neil, A. Q. & Tsaftaris, S. A., 15 Oct 2023, Machine Learning in Medical Imaging - 14th International Workshop, MLMI 2023, Held in Conjunction with MICCAI 2023, Proceedings. Cao, X., Ouyang, X., Xu, X., Rekik, I. & Cui, Z. (eds.). Springer Science and Business Media Deutschland GmbH, p. 234-242 9 p. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); vol. 14349 LNCS).

    Research output: Chapter in Book/Report/Conference proceedingConference contribution

  • Debiasing Counterfactuals in the Presence of Spurious Correlations

    Kumar, A., Fathi, N., Mehta, R., Nichyporuk, B., Falet, J. P. R., Tsaftaris, S. & Arbel, T., 9 Oct 2023, Clinical Image-Based Procedures, Fairness of AI in Medical Imaging, and Ethical and Philosophical Issues in Medical Imaging - 12th International Workshop, CLIP 2023 1st International Workshop, FAIMI 2023 and 2nd International Workshop, EPIMI 2023, Proceedings. Wesarg, S., Oyarzun Laura, C., Puyol Antón, E., King, A. P., Baxter, J. S. H., Erdt, M., Drechsler, K., Freiman, M., Chen, Y., Rekik, I., Eagleson, R., Feragen, A., Cheplygina, V., Ganz-Benjaminsen, M., Ferrante, E., Glocker, B., Moyer, D. & Petersen, E. (eds.). Springer Science and Business Media Deutschland GmbH, p. 276-286 11 p. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); vol. 14242 LNCS).

    Research output: Chapter in Book/Report/Conference proceedingConference contribution