#differential-privacy

[ follow ]
#natural-language-processing #entity-recognition #large-language-models #data-privacy #user-privacy
#large-language-models
fromInfoQ
1 week ago
Miscellaneous

Google Introduces VaultGemma: An Experimental Differentially Private LLM

fromInfoQ
1 week ago
Miscellaneous

Google Introduces VaultGemma: An Experimental Differentially Private LLM

more#large-language-models
Artificial intelligence
fromZDNET
2 weeks ago

How Google's new AI model protects user privacy without sacrificing performance

VaultGemma uses sequence-level differential privacy to prevent LLM memorization of sensitive training data while preserving high-quality model outputs.
Artificial intelligence
fromTechzine Global
2 weeks ago

Google launches VaultGemma: privacy AI without compromising performance

VaultGemma is a 1B-parameter differentially private language model that preserves performance while preventing memorization or leakage of sensitive data and will be open source.
Artificial intelligence
fromodsc.medium.com
4 weeks ago

Tokenizing Text for LLMs, an AI Agent Dictionary, Optimizing Agentic Workflows, and AI for Robotics at ODSC West

AI for robotics integrates foundation models, autonomous navigation, manipulation, agentic workflows, tokenization, metadata protocols, differential-privacy synthetic data, and hands-on training including a robotics hackathon prize.
Privacy technologies
fromHackernoon
1 year ago

The Role of Boundary Objects in New Tech Adoption | HackerNoon

The adoption of differential privacy at the U.S. Census Bureau highlights the need for effective stakeholder mediation in algorithmic governance.
fromHackernoon
1 year ago

The Census Didn't Just Get Safer-It Got More Complex | HackerNoon

The Bureau's framing of differential privacy (DP) as a "modern" and "advanced" confidentiality method leads to an oversimplified narrative, masking its broader social implications.
Digital life
Privacy technologies
fromHackernoon
1 year ago

LLM Probabilities, Training Size, and Perturbation Thresholds in Entity Recognition | HackerNoon

The article explores the intersection of natural language processing and privacy, emphasizing Differential Privacy as a method to protect data.
The research identifies key indicators for evaluating privacy risks in NLP applications.
#natural-language-processing
more#natural-language-processing
Artificial intelligence
fromZDNET
5 months ago

How Apple plans to train its AI on your data without sacrificing your privacy

Apple leverages synthetic data and differential privacy to enhance its AI technology while protecting user privacy.
[ Load more ]