2024/08/01 by Florian Dietz, Dietz, Florian, Dietrich Klakow +1
Computer Science · Mathematics · #Arithmetic #Block (permutation group theory) #Combinatorics #Computer network #Computer science #FOS: Computer and information sciences #Generalization #Geochemistry and Geologic Mapping #Machine Learning (cs.LG) #Mathematics #Modular design #Parallel computing #Programming language #Routing (electronic design automation)
paper · pdf · doi:10.48550/arxiv.2408.00508
published in arXiv (Cornell University) (Cornell University)
openalex publication_date 2024/08/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
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.