References & Citations
Computer Science > Computation and Language
Title: Minimum Description Length Recurrent Neural Networks
(Submitted on 31 Oct 2021 (v1), last revised 31 Mar 2022 (this version, v4))
Abstract: We train neural networks to optimize a Minimum Description Length score, i.e., to balance between the complexity of the network and its accuracy at a task. We show that networks optimizing this objective function master tasks involving memory challenges and go beyond context-free languages. These learners master languages such as $a^nb^n$, $a^nb^nc^n$, $a^nb^{2n}$, $a^nb^mc^{n+m}$, and they perform addition. Moreover, they often do so with 100% accuracy. The networks are small, and their inner workings are transparent. We thus provide formal proofs that their perfect accuracy holds not only on a given test set, but for any input sequence. To our knowledge, no other connectionist model has been shown to capture the underlying grammars for these languages in full generality.
Submission history
From: Nur Lan [view email][v1] Sun, 31 Oct 2021 21:43:31 GMT (785kb,D)
[v2] Fri, 25 Mar 2022 16:38:35 GMT (563kb,D)
[v3] Wed, 30 Mar 2022 13:09:34 GMT (578kb,D)
[v4] Thu, 31 Mar 2022 10:50:33 GMT (579kb,D)
Link back to: arXiv, form interface, contact.