The Power of Hard Attention Transformers on Data Sequences: A Formal Language Theoretic Perspective
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_ | 1866915014741327872 |
|---|---|
| author | Bergsträßer, Pascal Köcher, Chris Lin, Anthony Widjaja Zetzsche, Georg |
| author_facet | Bergsträßer, Pascal Köcher, Chris Lin, Anthony Widjaja Zetzsche, Georg |
| contents | Formal language theory has recently been successfully employed to unravel the power of transformer encoders. This setting is primarily applicable in Natural Language Processing (NLP), as a token embedding function (where a bounded number of tokens is admitted) is first applied before feeding the input to the transformer. On certain kinds of data (e.g. time series), we want our transformers to be able to handle arbitrary input sequences of numbers (or tuples thereof) without a priori limiting the values of these numbers. In this paper, we initiate the study of the expressive power of transformer encoders on sequences of data (i.e. tuples of numbers). Our results indicate an increase in expressive power of hard attention transformers over data sequences, in stark contrast to the case of strings. In particular, we prove that Unique Hard Attention Transformers (UHAT) over inputs as data sequences no longer lie within the circuit complexity class $AC^0$ (even without positional encodings), unlike the case of string inputs, but are still within the complexity class $TC^0$ (even with positional encodings). Over strings, UHAT without positional encodings capture only regular languages. In contrast, we show that over data sequences UHAT can capture non-regular properties. Finally, we show that UHAT capture languages definable in an extension of linear temporal logic with unary numeric predicates and arithmetics. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2405_16166 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | The Power of Hard Attention Transformers on Data Sequences: A Formal Language Theoretic Perspective Bergsträßer, Pascal Köcher, Chris Lin, Anthony Widjaja Zetzsche, Georg Formal Languages and Automata Theory Formal language theory has recently been successfully employed to unravel the power of transformer encoders. This setting is primarily applicable in Natural Language Processing (NLP), as a token embedding function (where a bounded number of tokens is admitted) is first applied before feeding the input to the transformer. On certain kinds of data (e.g. time series), we want our transformers to be able to handle arbitrary input sequences of numbers (or tuples thereof) without a priori limiting the values of these numbers. In this paper, we initiate the study of the expressive power of transformer encoders on sequences of data (i.e. tuples of numbers). Our results indicate an increase in expressive power of hard attention transformers over data sequences, in stark contrast to the case of strings. In particular, we prove that Unique Hard Attention Transformers (UHAT) over inputs as data sequences no longer lie within the circuit complexity class $AC^0$ (even without positional encodings), unlike the case of string inputs, but are still within the complexity class $TC^0$ (even with positional encodings). Over strings, UHAT without positional encodings capture only regular languages. In contrast, we show that over data sequences UHAT can capture non-regular properties. Finally, we show that UHAT capture languages definable in an extension of linear temporal logic with unary numeric predicates and arithmetics. |
| title | The Power of Hard Attention Transformers on Data Sequences: A Formal Language Theoretic Perspective |
| topic | Formal Languages and Automata Theory |
| url | https://arxiv.org/abs/2405.16166 |