Transformer with Tree-order Encoding for Neural Program Generation

Research output: Contributions to collected editions/worksArticle in conference proceedingsResearch

Standard

Transformer with Tree-order Encoding for Neural Program Generation. / Thellmann, Klaudia-Doris; Stadler, Bernhard; Usbeck, Ricardo et al.
Conference XXX. 2022.

Research output: Contributions to collected editions/worksArticle in conference proceedingsResearch

Harvard

APA

Vancouver

Thellmann KD, Stadler B, Usbeck R, Lehmann J. Transformer with Tree-order Encoding for Neural Program Generation. In Conference XXX. 2022 doi: 10.48550/arXiv.2206.13354

Bibtex

@inbook{56b3c87266394178bbbbf7097e04d0af,
title = "Transformer with Tree-order Encoding for Neural Program Generation",
abstract = " While a considerable amount of semantic parsing approaches have employed RNN architectures for code generation tasks, there have been only few attempts to investigate the applicability of Transformers for this task. Including hierarchical information of the underlying programming language syntax has proven to be effective for code generation. Since the positional encoding of the Transformer can only represent positions in a flat sequence, we have extended the encoding scheme to allow the attention mechanism to also attend over hierarchical positions in the input. Furthermore, we have realized a decoder based on a restrictive grammar graph model to improve the generation accuracy and ensure the well-formedness of the generated code. While we did not surpass the state of the art, our findings suggest that employing a tree-based positional encoding in combination with a shared natural-language subword vocabulary improves generation performance over sequential positional encodings. ",
keywords = "cs.CL, cs.AI, 68T07, 68T50, I.2.7, Informatics",
author = "Klaudia-Doris Thellmann and Bernhard Stadler and Ricardo Usbeck and Jens Lehmann",
note = "This paper was authored in late 2020 and early 2021 for the most part",
year = "2022",
month = may,
day = "30",
doi = "10.48550/arXiv.2206.13354",
language = "English",
booktitle = "Conference XXX",

}

RIS

TY - CHAP

T1 - Transformer with Tree-order Encoding for Neural Program Generation

AU - Thellmann, Klaudia-Doris

AU - Stadler, Bernhard

AU - Usbeck, Ricardo

AU - Lehmann, Jens

N1 - This paper was authored in late 2020 and early 2021 for the most part

PY - 2022/5/30

Y1 - 2022/5/30

N2 - While a considerable amount of semantic parsing approaches have employed RNN architectures for code generation tasks, there have been only few attempts to investigate the applicability of Transformers for this task. Including hierarchical information of the underlying programming language syntax has proven to be effective for code generation. Since the positional encoding of the Transformer can only represent positions in a flat sequence, we have extended the encoding scheme to allow the attention mechanism to also attend over hierarchical positions in the input. Furthermore, we have realized a decoder based on a restrictive grammar graph model to improve the generation accuracy and ensure the well-formedness of the generated code. While we did not surpass the state of the art, our findings suggest that employing a tree-based positional encoding in combination with a shared natural-language subword vocabulary improves generation performance over sequential positional encodings.

AB - While a considerable amount of semantic parsing approaches have employed RNN architectures for code generation tasks, there have been only few attempts to investigate the applicability of Transformers for this task. Including hierarchical information of the underlying programming language syntax has proven to be effective for code generation. Since the positional encoding of the Transformer can only represent positions in a flat sequence, we have extended the encoding scheme to allow the attention mechanism to also attend over hierarchical positions in the input. Furthermore, we have realized a decoder based on a restrictive grammar graph model to improve the generation accuracy and ensure the well-formedness of the generated code. While we did not surpass the state of the art, our findings suggest that employing a tree-based positional encoding in combination with a shared natural-language subword vocabulary improves generation performance over sequential positional encodings.

KW - cs.CL

KW - cs.AI

KW - 68T07, 68T50

KW - I.2.7

KW - Informatics

U2 - 10.48550/arXiv.2206.13354

DO - 10.48550/arXiv.2206.13354

M3 - Article in conference proceedings

BT - Conference XXX

ER -

Recently viewed

Activities

  1. Taking ICALL to task: Blending form-focus & task-based foreign language learning
  2. Coherent sets in nonautonomous dynamics
  3. Memory Acts: Memory without Representation.: Theoretical and Methodological Suggestions
  4. Alterations of a visual and how they work for and at the boundaries of an interorganizational team: A multimodal exploration
  5. The golden age of software architecture better named the middle age of software architecture - Some provocative thoughts
  6. Judgement Practices in the Artistic Field
  7. Princeton University
  8. The global classroom. Introduction, presenation and workshop: Introduction, presenation and workshop
  9. Activating an Integrative Mindset Improves the Subjective Outcomes of Value-Driven Conflicts
  10. 13th IFAC and IEEE Conference on Programmable Devices and Embedded Systems - PDeS 2015
  11. Workshop of the Nordic Research Network in Memory Studies - 2013
  12. Plasma shock wave simulation for laser shock processing
  13. Workshop: The Black Technical Object
  14. Modelling biodegradability based on OECD 301D data for the design of mineralising ionic liquids
  15. Transdisciplinary Evaluation of Different Coastal Adaptation Strategies: Integrating Regional Perceptions of Scientists, Practitioners and the Public
  16. Interpretation and contestation of fracking in a changing context: The case of Germany and its proclaimed energy transition
  17. Reading strategy instruction and students' perceptions on fostering self-regulated reading
  18. 4th CRITICS Workshop and Winter School on “Critical Transitions in Complex Systems: Mathematical theory and applications”
  19. Interpreting integrative political strategies as boundary objects
  20. Interactions between social movements and international organisations

Publications

  1. A geometric approach for controlling an electromagnetic actuator with the help of a linear Model Predictive Control
  2. Modeling Conditional Dependencies in Multiagent Trajectories
  3. For a return to the forgotten formula: 'Data 1 + Data 2 > Data 1'
  4. An evaluation of BPR methodologies adopting NIMSAD: A systematic framework for understanding and evaluating methodologies
  5. A PHENOMENOGRAPHICAL STUDY OF CHILDRENS’ SPATIAL THOUGHT WHILE USING MAPS IN REAL SPACES
  6. Age effects on controlling tools with sensorimotor transformations
  7. A cascade controller structure using an internal PID controller for a hybrid piezo-hydraulic actuator in camless internal combustion engines
  8. Evaluating structural and compositional canopy characteristics to predict the light-demand signature of the forest understorey in mixed, semi-natural temperate forests
  9. Automatic feature selection for anomaly detection
  10. Performance incentives in activity-based management
  11. Comparison of Bio-Inspired Algorithms in a Case Study for Optimizing Capacitor Bank Allocation in Electrical Power Distribution
  12. Tuning kalman filter in linear systems
  13. How mobile app design impacts user responses to mixed self-tracking outcomes
  14. Implementation of formative assessment
  15. A Comparative Study for Fisheye Image Classification
  16. The role of task meaning on output in groups
  17. Optimal scheduling of AGVs in a reentrant blocking job-shop
  18. General Patterns and Conclusions
  19. Model-based nonlinear filter design for tower load reduction of wind power plants with active power control capability
  20. Self-perceived quality of life predicts mortality risk better than a multi-biomarker panel, but the combination of both does best
  21. Comparison of Supervised versus Self-Administered Stretching on Bench Press Maximal Strength and Force Development
  22. Putting adaptive planning into practice: A meta-analysis of current applications
  23. Almost-invariant and finite-time coherent sets
  24. A common European asylum system? How variation in Member States’ administrative capacity undermines EU asylum harmonisation
  25. Studying properties of water data using manifold-aware anomaly detectors
  26. Algorithmisches Management
  27. Quantification of amino acids in fermentation media by isocratic HPLC analysis of their