“Intriguing Properties of Generative Classifiers”, 2023-09-28 ():
[Twitter] What is the best paradigm to recognize objects—discriminative inference (fast but potentially prone to shortcut learning) or using a generative model (slow but potentially more robust)?
We build on recent advances in generative modeling that turn text-to-image models into classifiers. This allows us to study their behavior and to compare them against discriminative models and human psychophysical data.
We report 4 intriguing emergent properties of generative classifiers: they show a record-breaking human-like shape bias (99% for Imagen), near human-level out-of-distribution accuracy, state-of-the-art alignment with human classification errors, and they understand certain perceptual illusions.
Our results indicate that while the current dominant paradigm for modeling human object recognition is discriminative inference, zero-shot generative models approximate human object recognition data surprisingly well.