Block-Operations: Using Modular Routing to Improve Compositional Generalization
Fuente:
arXiv
Salvato in:
| Autori principali: | , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866929445975097344 |
|---|---|
| author | Dietz, Florian Klakow, Dietrich |
| author_facet | Dietz, Florian Klakow, Dietrich |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2408_00508 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Block-Operations: Using Modular Routing to Improve Compositional Generalization Dietz, Florian Klakow, Dietrich Machine Learning 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. |
| title | Block-Operations: Using Modular Routing to Improve Compositional Generalization |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2408.00508 |