Data Anonymization for Telco AI Use Cases – Sridhar Rao, The Linux Foundation
Data Anonymization for Telco AI Use Cases – Sridhar Rao, The Linux Foundation
Data anonymization can help Telcos to share data to facilitate open innovation. Two big challenges to address while anonymizing Telco data for AI usecases are (a) fool-proof against de-anonymization (b) Not hamper the power (Ex: predictive, Classification) of the AI models. The talk we cover the following:
(a) State of art of Data Anonymization applied to Telcos, including the research works, projects and specifications.
(b) What constitutes the sensitive data (names, addresses, telco-specific fields, location-data, etc). in Telco scenarios
(c) Anonymization categories such as (Suppression, Masking, Pseudonymization, Generalization, Swapping, Perturbation and Synthetic Data Generation).
(d) Approaches/techniques ranging from classic (ex: K-Anonymity) to use of NLP to GANs (Generative Adversarial Networks).
(e) Which of the above categories and techniques are applicable to Telco Data, considering the challenges of deanonymization and model-power.
(f) Demonstration of the developed unified-anonymization tool.
The talk will also include description of the libraries that can be used, demonstration of the techniques, and showcasing of the impact of anonymization on the AI models.
by LF Networking
linux foundation
