Newest papers

Hierarchical Token Merging with Recursive Autoencoders for Unsupervised Grammar Induction

Murat Guney Kemal, Robert Lieck · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26695 · cited by 0 works (OpenAlex) · open-access PDF

Indonesian

Pembelajaran hierarki secara langsung dari data, tanpa pengawasan struktural eksplisit, tetap menjadi tantangan sentral dalam pembelajaran representasi. Kami menyelidiki apakah struktur hierarki yang bermakna dapat muncul dari kompresi saja. Sementara autoencoders mempelajari representasi yang dikompresi melalui rekonstruksi, tujuan ini tidak semestinya mendukung struktur komposisi yang bermakna. Kami mempelajari tantangan ini melalui induction grammar yang tidak diawasi, di mana hierarki yang mendasarinya dapat dievaluasi secara eksplisit. Kami memperkenalkan sebuah hierarki token-merging autoencoding yang berulang kali menggabungkan token-token yang berdekatan menjadi representasi yang semakin abstrak, mengkompresi seluruh struktur komposisi melalui hierarki variasi yang kuat. Model ini bersama-sama mempelajari representasi dan struktur hierarki dengan membandingkan representasi dan struktur hierarki dengan

Machine translation (Meta M2M100); the English original follows.

Abstract

Hierarchical abstraction is a fundamental principle across many data modalities, enabling complex sequences to be represented compositionally at multiple levels. Learning such hierarchies directly from data, without explicit structural supervision, remains a central challenge in representation learning. We investigate whether meaningful hierarchical structure can emerge from compression alone. While autoencoders learn compressed representations through reconstruction, this objective does not necessarily favour meaningful compositional structure. We study this challenge through unsupervised grammar induction, where the underlying hierarchy can be evaluated explicitly. We introduce a hierarchical token-merging autoencoder that recursively combines adjacent tokens into increasingly abstract representations, compressing entire sequences through a variational information bottleneck. The model jointly learns representations and hierarchical structure by marginalising over possible binary trees using dynamic programming. To address training instability, we combine independently trained models through unsupervised consensus and distillation. Experiments on a synthetic, ambiguous grammar demonstrate the emergence of meaningful hierarchical structure without structural supervision. The resulting model achieves 93% grammatical parse-validity and 93% exact reconstruction on held-out sequences, compared with 42% parse-validity for the strongest trivial baseline. These findings establish a proof of concept for hierarchical token merging as a mechanism for unsupervised structure discovery and abstraction learning.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Hierarchy, Autoencoder, Artificial intelligence, Abstraction, Representation (politics)

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange