Timezone: »

MARS: Meta-learning as score matching in the function space
Kruno Lehman · Jonas Rothfuss · Andreas Krause
Event URL: https://openreview.net/forum?id=hV0YobaZ0xp »

We approach meta-learning through the lens of functional Bayesian neural network inference which views the prior as a stochastic process and performs inference in the function space. Specifically, we view the meta-training tasks as samples from the data-generating process and formalize meta-learning as empirically estimating the law of this stochastic process. Our approach can seamlessly acquire and represent complex prior knowledge by meta-learning the score function of the data-generating process marginals. In a comprehensive benchmark, we demonstrate that our method achieves state-of-the-art performance in terms of predictive accuracy and substantial improvements in the quality of uncertainty estimates.

Author Information

Kruno Lehman (Inria / ETH Zürich)

• Researcher at Inria • MSc student in Statistics at ETH Zürich • Predoctoral research intern at University of Oxford • Member of QueerInAI

Jonas Rothfuss (ETH Zurich)
Andreas Krause (ETH Zurich)

More from the Same Authors