Filters








18 Hits in 0.83 sec

Streamable Regular Transductions [article]

Rajeev Alur, Dana Fisman, Konstantinos Mamouras, Mukund Raghothaman, Caleb Stanford
2019 arXiv   pre-print
These machines give rise to the class of streamable regular transductions (SR), and to the class of streamable linear regular transductions (SLR) when the register updates are copyless, i.e. every register  ...  We consider the model of unambiguous Cost Register Automata (CRAs), which are machines that combine finite-state control (for identifying regular patterns) with a finite set of data registers (for computing  ...  We define the class of streamable regular (resp., streamable linear regular ) transductions, denoted SR (resp., SLR), as the class of transductions that are computed by copyful (resp., copyless) unambiguous  ... 
arXiv:1807.03865v2 fatcat:5tcdnsssgjdcjjarpemlecfsim

Conv-Transformer Transducer: Low Latency, Low Frame Rate, Streamable End-to-End Speech Recognition

Wenyong Huang, Wenchao Hu, Yu Ting Yeung, Xiao Chen
2020 Interspeech 2020  
To make Transformer suitable for streaming ASR, we explore Transducer framework as a streamable way to learn alignments.  ...  The performance is comparable to previously published streamable Transformer Transducer and strong hybrid streaming ASR systems, and is achieved with smaller look-ahead window (140 ms), fewer parameters  ...  Conv-Transformer Transducer Transducer Transducer, proposed in [10] , is a neural sequence transduction framework which can be used for end-to-end speech recognition, i.e., transduction of an input  ... 
doi:10.21437/interspeech.2020-2361 dblp:conf/interspeech/HuangHYC20 fatcat:ybivox2fkncwpnewgbssucslem

Conv-Transformer Transducer: Low Latency, Low Frame Rate, Streamable End-to-End Speech Recognition [article]

Wenyong Huang, Wenchao Hu, Yu Ting Yeung, Xiao Chen
2020 arXiv   pre-print
To make Transformer suitable for streaming ASR, we explore Transducer framework as a streamable way to learn alignments.  ...  The performance is comparable to previously published streamable Transformer Transducer and strong hybrid streaming ASR systems, and is achieved with smaller look-ahead window (140~ms), fewer parameters  ...  Conv-Transformer Transducer Transducer Transducer, proposed in [10] , is a neural sequence transduction framework which can be used for end-to-end speech recognition, i.e., transduction of an input  ... 
arXiv:2008.05750v1 fatcat:rdyubwka3beprhjzfrtgfvqtsi

Transformer Transducer: A Streamable Speech Recognition Model with Transformer Encoders and RNN-T Loss [article]

Qian Zhang, Han Lu, Hasim Sak, Anshuman Tripathi, Erik McDermott, Stephen Koo, Shankar Kumar
2020 arXiv   pre-print
), LAS (e2e), RNN-T (e2e & streamable) and Transformer Transducer models (e2e & streamable) on LibriSpeech test sets.  ...  Again, dropout to both dense layers for regularization, and a residual connection of normalized input and the output of the second dense layer (i.e.  ... 
arXiv:2002.02562v2 fatcat:n7zabn3rubav7ce4nb5y6puez4

Streamability of nested word transductions [article]

Emmanuel Filiot and Olivier Gauwin and Pierre-Alain Reynier and Frédéric Servais
2019 arXiv   pre-print
., in a single left-to-right pass) a nested word transduction with a limited amount of memory.  ...  This condition defines a class of transductions that strictly contains all determinizable VPTs.  ...  Vol. 15:2 STREAMABILITY OF NESTED WORD TRANSDUCTIONS 1:11 1 : 12 : E.Filiot, O. Gauwin, P.-A. Reynier, and F.  ... 
arXiv:1707.00527v3 fatcat:3ojt73s3njfc3cgtkux7mc52zu

From Two-Way to One-Way Finite State Transducers [article]

Emmanuel Filiot and Olivier Gauwin and Pierre-Alain Reynier and Frédéric Servais
2013 arXiv   pre-print
This well-known result, shown by Rabin and Scott and independently by Shepherdson, states that two-way finite state automata (even non-deterministic) characterize the class of regular languages.  ...  By extending Rabin and Scott's proof to transductions, we show that this problem is decidable.  ...  Then M q1,q2 is regular.  ... 
arXiv:1301.5197v2 fatcat:egiwbcebbjgyxl7scxa5t5dax4

Streamability of nested word transductions

Emmanuel Filiot, Olivier Gauwin, Pierre-Alain Reynier, Frédéric Servais
2019 Logical Methods in Computer Science  
., in a single left-to-right pass) a nested word transduction with a limited amount of memory.  ...  This condition defines a class of transductions that strictly contains all determinizable VPTs.  ...  nested words, deterministic VPTs are too restrictive to capture all streamable VPT transformations.  ... 
doi:10.23638/lmcs-15(2:1)2019 fatcat:h7wvw5rwungdtaw46ai7s3ijoe

Streamability of Nested Word Transductions

Emmanuel Filiot, Olivier Gauwin, Pierre-Alain Reynier, Frédéric Servais, Marc Herbstritt
2011 Foundations of Software Technology and Theoretical Computer Science  
This condition defines a class of transductions that strictly contains all determinizable VPTs.  ...  We show that it is decidable in coNPTime for a nested word transduction defined by a visibly pushdown transducer (VPT), if it is HBM.  ...  Moreover, BM is undecidable for pushdown transducers, since it is as difficult as deciding whether a pushdown automaton defines a regular language.  ... 
doi:10.4230/lipics.fsttcs.2011.312 dblp:conf/fsttcs/FiliotGRS11 fatcat:zsxbkypta5eefkwbzdq6isrmhq

Frontmatter, Table of Contents, Preface, Conference Organization, External Reviewers

Supratik Chakraborty, Amit Kumar, Marc Herbstritt
2011 Foundations of Software Technology and Theoretical Computer Science  
Streamability of Nested Word Transductions Emmanuel Filiot, Olivier Gauwin, Pierre-Alain Reynier, and Frédéric Servais . . . .  ...  Parameterized Regular Expressions and Their Languages Pablo Barceló, Leonid Libkin, and Juan L. Reutter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .  ... 
doi:10.4230/lipics.fsttcs.2011.i dblp:conf/fsttcs/ChakrabortyK11 fatcat:mabhkcxew5axbfxfj3t6ob5kei

Tree Transducers and Formal Methods (Dagstuhl Seminar 13192)

Sebastian Maneth, Helmut Seidl, Marc Herbstritt
2013 Dagstuhl Reports  
Many transformations such as flips of document pairs are hyperstreamable but not streamable.  ...  By extending Rabin and Scott's proof to transductions, we show that this problem is decidable.  ... 
doi:10.4230/dagrep.3.5.1 dblp:journals/dagstuhl-reports/ManethS13 fatcat:xxotqpzgnzc3zdlkk4zxsdot5i

Visibly pushdown transducers

Emmanuel Filiot, Jean-François Raskin, Pierre-Alain Reynier, Frédéric Servais, Jean-Marc Talbot
2018 Journal of computer and system sciences (Print)  
or moving subtrees) makes the transformations they define more likely to be non streamable.  ...  NPA) define the class of regular languages (resp. context-free languages). We denote by CFL the class of context-free languages. Transductions. Let Σ and ∆ be two alphabets.  ... 
doi:10.1016/j.jcss.2018.05.002 fatcat:yqiq2kfuwrhvzlnldlpb2qwsii

Equivalence of Extended Symbolic Finite Transducers [chapter]

Loris D'Antoni, Margus Veanes
2013 Lecture Notes in Computer Science  
These representations are either not succinct or not streamable.  ...  Introduction Finite automata have proven to be an effective tool in a wide range of applications, from regular expressions to network packet inspection [15] .  ... 
doi:10.1007/978-3-642-39799-8_41 fatcat:3kdaqeg6dfhi3ff7pmf6njvzb4

Dual-mode ASR: Unify and Improve Streaming ASR with Full-context Modeling [article]

Jiahui Yu, Wei Han, Anmol Gulati, Chung-Cheng Chiu, Bo Li, Tara N. Sainath, Yonghui Wu, Ruoming Pang
2021 arXiv   pre-print
DUAL-MODE ASR Most neural sequence transduction networks for ASR have an encoder-decoder structure (Graves, 2012; Li et al., 2020a) , as shown in Figure 1 .  ...  Compared with sequence-to-sequence models (Chorowski et al., 2014; Chorowski & Jaitly, 2016; Bahdanau et al., 2016; Chan et al., 2016) , RNN-T models are naturally streamable and have shown great potentials  ... 
arXiv:2010.06030v2 fatcat:hiwfxvkvafh2tcfwlqcbknpo4e

Unported license Tree Transducers and Formal Methods

Sebastian Maneth, Helmut Seidl, Sebastian Maneth, Sebastian Maneth, Helmut Seidl, Dagstuhl Reports
Dagstuhl Reports   unpublished
Many transformations such as flips of document pairs are hyperstreamable but not streamable.  ...  By extending Rabin and Scott's proof to transductions, we show that this problem is decidable.  ... 
fatcat:hbovpv236zezfesbns6zomyfta

Visibly Pushdown Transducers [chapter]

Jean-François Raskin, Frédéric Servais
Lecture Notes in Computer Science  
or moving subtrees) makes the transformations they define more likely to be non streamable.  ...  NPA) define the class of regular languages (resp. context-free languages). We denote by CFL the class of context-free languages. Transductions. Let Σ and ∆ be two alphabets.  ... 
doi:10.1007/978-3-540-70583-3_32 fatcat:znclapdfenaezeg3psudu7t3ki
« Previous Showing results 1 — 15 out of 18 results