Compositional Generalization via Forced Rendering of Disentangled Latents
This addresses a key challenge in AI for building models that generalize compositionally, though it is incremental as it builds on existing disentanglement research with controlled experiments.
The paper tackles the problem of compositional generalization in deep learning by showing that standard generative models fail to compose disentangled factors for out-of-distribution samples due to re-entangling in later layers, and demonstrates that forcing models to render disentangled latents directly into output space enables highly data-efficient and effective OOD composition.
Composition-the ability to generate myriad variations from finite means-is believed to underlie powerful generalization. However, compositional generalization remains a key challenge for deep learning. A widely held assumption is that learning disentangled (factorized) representations naturally supports this kind of extrapolation. Yet, empirical results are mixed, with many generative models failing to recognize and compose factors to generate out-of-distribution (OOD) samples. In this work, we investigate a controlled 2D Gaussian "bump" generation task with fully disentangled (x,y) inputs, demonstrating that standard generative architectures still fail in OOD regions when training with partial data, by re-entangling latent representations in subsequent layers. By examining the model's learned kernels and manifold geometry, we show that this failure reflects a "memorization" strategy for generation via data superposition rather than via composition of the true factorized features. We show that when models are forced-through architectural modifications with regularization or curated training data-to render the disentangled latents into the full-dimensional representational (pixel) space, they can be highly data-efficient and effective at composing in OOD regions. These findings underscore that disentangled latents in an abstract representation are insufficient and show that if models can represent disentangled factors directly in the output representational space, it can achieve robust compositional generalization.