Synthetic Data 101

Video: https://video.linux.it/w/6HoHhhyrnfqYnzZiM9sRx5?start=42m54s

Per sfruttare appieno il potenziale dell'intelligenza artificiale, è necessario disporre di una vasta quantità di dati. Tuttavia, ottenere e utilizzare tali dati non è un compito semplice. Esaminiamo insieme come è possibile generare dati tramite gli algoritmi di intelligenza artificiale utilizzando dati reali, ricreando situazioni di vita reale in modo matematico e statistico.

L'avvio di progetti di intelligenza artificiale può essere difficile per diverse ragioni, tra cui l'organizzazione, le aspettative e i costi. Tuttavia, uno dei principali ostacoli è rappresentato dai dati stessi. Ottenere i dati richiesti può essere complesso e costoso, e spesso le informazioni contenute sono sensibili, rendendole inaccessibili a causa delle norme sulla privacy. In questo contesto, i Dati Sintetici possono svolgere un ruolo importante, consentendoci di ottenere dati simili a quelli reali, ma senza problemi di privacy e con costi più contenuti.

Luca Gilli — Direttore tecnico di Clearbox AI Solutions


Transcript

1. Synthetic Data 101 Luca Gilli Torino, 18 Ottobre 2023
2. 2 • Co-founded Clearbox AI, a synthetic data startup hosted by the university incubator of Politecnico di Torino • I have a background in applied mathematics and scientific software development • Lived in the Netherlands for 10 years before moving to Val di Susa About myself
3. Gartner estimates that by 2030, synthetic data will completely overshadow real data in AI models. Gartner, Is Synthetic Data the Future of AI?, 2022 3
4. 4 Synthetic data Definition A synthetic dataset is obtained by generating fictitious data that incorporates the statistical properties and distributions of an original dataset, thus resulting realistic.
5. 5 Why the hype: Access and Quality Data Privacy • Privacy issues related to data sharing (GDPR/CCPA) New Anonymization paradigm 01 02 Data Augmentation • Class imbalance and ML models generalization issues It improves models’ performances
6. 6 Why the hype: Access and Quality Data Privacy • Privacy issues related to data sharing (GDPR/CCPA) New Anonymization paradigm 01 02 Data Augmentation • Class imbalance and ML models generalization issues It improves models’ performances
7. 7 Why the hype: Access and Quality Data Privacy • Privacy issues related to data sharing (GDPR/CCPA) New Anonymization paradigm 01 https://edps.europa.eu/press-publications/publications/techsonar/synth etic-data_en
8. Synthetic data as a Privacy Enhancing Technology 8
9. Anonymisation !=Pseudonymisation Anonymising behavioural data is really challenging. 9
10. http://www.tdp.cat/issues16/tdp.a363a19.pdf 10
11. http://www.tdp.cat/issues16/tdp.a363a19.pdf 11
12. https://edps.europa.eu/press-publications/publications/techsonar/synth etic-data_en 12
13. How to generate synthetic data 13
14. Synthetic Data Generation
15. Synthetic Data Generation https://lilianweng.github.io/posts/2021-07-11-diffusion-models/ Several families of generative models. GANs gained popularity in 2017, Diffusion models are becoming state-of-the-art.
16. 16 Synthetic Data Generation
17. El Emam, K., “Practical Synthetic Data Generation”, O’Reilly Synthetic Data Generation 17
18. El Emam, K., “Practical Synthetic Data Generation”, O’Reilly Synthetic Data Generation 18 Measuring the information contained within the synthetic data. Measuring the ‘novelty’ of the synthetic dataset.
19. In conclusion Synthetic data generators are a modern and powerful Privacy Enhancing Technology. Trust in synthetic data → generating synthetic data goes beyond using generative models but requires extensive privacy and quality tests. 19
20. @ClearboxAI www.clearbox.ai info@clearbox.ai Thanks for listening! Feel free to contact us: