Timezone: »

 
BatchGFN: Generative Flow Networks for Batch Active Learning
Shreshth Malik · Salem Lahlou · Andrew Jesson · Moksh Jain · Nikolay Malkin · Tristan Deleu · Yoshua Bengio · Yarin Gal

@
in Structured Probabilistic Inference and Generative Modeling »
Event URL: https://openreview.net/forum?id=U0aM6wrpFR »

We introduce BatchGFN—a novel approach for pool-based active learning that uses generative flow networks to sample sets of data points proportional to a batch reward. With an appropriate reward function to quantify the utility of acquiring a batch, such as the joint mutual information between the batch and the model parameters, BatchGFN is able to construct highly informative batches for active learning. We show our approach enables sampling near-optimal utility batches at inference time with a single forward pass per point in the batch in toy regression problems. This alleviates the computational complexity of batch-aware algorithms and removes the need for greedy approximations to find maximizers for the batch reward. We also present early results for amortizing training across acquisition steps, which will enable scaling to real-world tasks.

Keywords: Active Learning · Generative Flow Networks · Batch Active Learning · Deep Generative Modelling · Probabilistic Modelling

Author Information

Shreshth Malik (University of Oxford)
Salem Lahlou (Mila, Université de Montréal)
Andrew Jesson (University of Oxford)
Moksh Jain (Mila / Université de Montréal)
Nikolay Malkin (Mila / Université de Montréal)
Tristan Deleu (Mila - Université de Montréal)
Yoshua Bengio (Mila - Quebec AI Institute)
Yarin Gal (University of Oxford)

More from the Same Authors