"Towards Value-Centric Big Data" e-SIDES Workshop - "Privacy Preserving Technologies - An outlook", Tjerk Timan
0 likes80 views
The document outlines the discussions from the E-Sides conference on privacy-preserving technologies and regulatory challenges in the context of data processing, particularly for EU citizens. It highlights technologies like self-sovereign identity, multi-party computation, and explainable AI as potential solutions to issues of data control, transparency, and privacy. Key topics include the balancing of human values with big data analytics, and the need for continuous adaptation of policies in response to new technological challenges.
"Towards Value-Centric Big Data" e-SIDES Workshop - "Privacy Preserving Technologies - An outlook", Tjerk Timan
- 2. Data Single Market - regulatory challenges Data Processing data Applying data-driven innovations personal non-personal AI and ML, cloud computing data protection for EU citizens free flow of data and geo-blocking IP and copyright, data-, and AI marketplaces, cybersecurity & access novel business propositions, value creation, lean policy algorithmic transparency, inclusiveness for the BDVA - support via ‘how to’s’ and best practices (if any) on most pressing topics (GDRP implementation for startups for example) BDVe context
- 3. What? 1. an overview of technologies to inform policymakers 2. an aligning of research agendas 3. a discussion on future challenges
- 6. What is new? 1. From big data to data4AI 2. Geographical scope (where is big data?) 3. Conceptual scope (what is (personal) data?) 4. Socio-politcal scope (why is it relevant?)
- 9. Sticky policies https://www.hpl.hp.com/techreports/2003/HPL-2003-49.pdf What does it do? codifies policies and sends it along with a dataset What does it solve? Enforcement of regulation New Problem? Needs a trusted party - needs updates
- 10. Self-Sovereign Identity What does it do? Turns data control back to user What does it solve? Power balance and lack of data control New Problem? Cognitive load - assumption that control is positive
- 11. Multi-Party Computation What does it do? makes possible to analyse big data without seeing it What does it solve? Anonimisation and data sharing - third party problem New Problem? Performance - only works on aggregatate level
- 12. Explainaible AI What does it do? Explains decisions made by ML What does it solve? Inherent intransparencey of DL/ML New Problem? Recursiveness: How transparent are XAI outcomes?
- 15. PPT maturity Societal uptake PPT technical factors: - experVse needed to implement - effect or performance - compaVbility - updatability - standardised vs customised societal factors - role of privacy in society - plaaorm dominance - understanding of data - use/role of big data and AI - level of trust in digital tech
- 16. Some questions for the afternoon 1. What is data4AI that big data is not? 2. What can we say about privacy&AI? 3. Should we still talk about privacy? 4. What are the limits of techno regulation - do we have choice?