Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore principled approaches to measuring memorization in foundation models through this 44-minute workshop presentation by Kamalika Chaudhuri from UC San Diego, delivered as part of the IFDS Workshop series at the Paul G. Allen School. Examine the critical challenge of understanding how large-scale AI models memorize training data and learn systematic methodologies for quantifying this phenomenon. Discover theoretical frameworks and practical techniques for assessing memorization patterns in foundation models, including methods for distinguishing between genuine learning and rote memorization. Gain insights into the implications of memorization for model generalization, privacy concerns, and the development of more robust AI systems. Understand the current state of research in this rapidly evolving field and explore potential solutions for creating more principled approaches to memorization measurement in modern machine learning architectures.
