Efficient Iterative Amortized Inference for Learning Symmetric and Disentangled Multi-Object Representations
This work addresses the problem of slow and memory-intensive learning of object-centric representations for AI researchers, offering a more practical solution with significant speed improvements.
The paper tackles the challenge of inefficient training and inference in unsupervised multi-object representation learning by introducing EfficientMORL, which reduces iterative amortized inference steps through a two-stage approach, achieving 99.1% performance with zero refinement steps at test time and nearly 10x faster training and inference than prior state-of-the-art.
Unsupervised multi-object representation learning depends on inductive biases to guide the discovery of object-centric representations that generalize. However, we observe that methods for learning these representations are either impractical due to long training times and large memory consumption or forego key inductive biases. In this work, we introduce EfficientMORL, an efficient framework for the unsupervised learning of object-centric representations. We show that optimization challenges caused by requiring both symmetry and disentanglement can in fact be addressed by high-cost iterative amortized inference by designing the framework to minimize its dependence on it. We take a two-stage approach to inference: first, a hierarchical variational autoencoder extracts symmetric and disentangled representations through bottom-up inference, and second, a lightweight network refines the representations with top-down feedback. The number of refinement steps taken during training is reduced following a curriculum, so that at test time with zero steps the model achieves 99.1% of the refined decomposition performance. We demonstrate strong object decomposition and disentanglement on the standard multi-object benchmark while achieving nearly an order of magnitude faster training and test time inference over the previous state-of-the-art model.