Block-Operations: Using Modular Routing to Improve Compositional Generalization
This addresses a fundamental problem in neural network generalization for AI researchers, though it is incremental as it builds on existing architectures.
The paper tackles poor compositional generalization in neural networks by proposing block-operations and the Multiplexer component, which splits activations into blocks to encourage modular routing, resulting in the model learning underlying processes on synthetic and realistic tasks where FNNs and Transformers failed.
We explore the hypothesis that poor compositional generalization in neural networks is caused by difficulties with learning effective routing. To solve this problem, we propose the concept of block-operations, which is based on splitting all activation tensors in the network into uniformly sized blocks and using an inductive bias to encourage modular routing and modification of these blocks. Based on this concept we introduce the Multiplexer, a new architectural component that enhances the Feed Forward Neural Network (FNN). We experimentally confirm that Multiplexers exhibit strong compositional generalization. On both a synthetic and a realistic task our model was able to learn the underlying process behind the task, whereas both FNNs and Transformers were only able to learn heuristic approximations. We propose as future work to use the principles of block-operations to improve other existing architectures.