Filters








3,189 Hits in 8.3 sec

Overfitting in Synthesis: Theory and Practice (Extended Version) [article]

Saswat Padhi, Todd Millstein, Aditya Nori, Rahul Sharma
2019 arXiv   pre-print
A standard approach to mitigate overfitting in machine learning is to run multiple learners with varying expressiveness in parallel.  ...  In syntax-guided synthesis (SyGuS), a synthesizer's goal is to automatically generate a program belonging to a grammar of possible implementations that meets a logical specification.  ...  This work was supported in part by the National Science Foundation (NSF) under grants CCF-and CCF-. The lead author was also supported by an internship and a PhD Fellowship from Microsoft Research.  ... 
arXiv:1905.07457v3 fatcat:yx5vqaecubdf5oceg5ftxmj47u

Overfitting in Synthesis: Theory and Practice [chapter]

Saswat Padhi, Todd Millstein, Aditya Nori, Rahul Sharma
2019 Lecture Notes in Computer Science  
A standard approach to mitigate overfitting in machine learning is to run multiple learners with varying expressiveness in parallel.  ...  In syntax-guided synthesis (SyGuS), a synthesizer's goal is to automatically generate a program belonging to a grammar of possible implementations that meets a logical specification.  ...  This work was supported in part by the National Science Foundation (NSF) under grants CCF-1527923 and CCF-1837129.  ... 
doi:10.1007/978-3-030-25540-4_17 fatcat:vx462tsefraydnwm7c2tkti24i

Automated Transpilation of Imperative to Functional Code using Neural-Guided Program Synthesis (Extended Version) [article]

Benjamin Mariano, Yanju Chen, Yu Feng, Greg Durrett, Isil Dillig
2022 arXiv   pre-print
We have implemented our approach in a tool called NGST2 and use it to translate imperative Java and Python code to functional variants that use the Stream and functools APIs respectively.  ...  Our method leverages this observation to design a new neural-guided synthesis algorithm that (1) uses a novel neural architecture called cognate grammar network (CGN) and (2) leverages a form of concolic  ...  1:25 grant numbers CCF-1908494, CCF-1811865, CCF-1918889, DARPA under the HARDEN program, Google under the Google Faculty Research Grant, as well as both Intel and RelationalAI.  ... 
arXiv:2203.09452v2 fatcat:cff7ogv5wfgwlfnql6kvzhpngm

Overfitting in semantics-based automated program repair

Xuan Bach D. Le, Ferdian Thung, David Lo, Claire Le Goues
2018 Empirical Software Engineering  
the nature of overfitting in semantics-based APR.  ...  Semantics-based APR uses symbolic execution and test suites to extract semantic constraints, and uses program synthesis to synthesize repairs that satisfy the extracted constraints.  ...  We also thank the authors of CVC4 and Enumerative synthesis engines for making these engines accessible. Empir Software Eng  ... 
doi:10.1007/s10664-017-9577-2 fatcat:t52pday2kjc3vdgktv42c2ozb4

Learning Nonlinear Loop Invariants with Gated Continuous Logic Networks (Extended Version) [article]

Jianan Yao, Gabriel Ryan, Justin Wong, Suman Jana, Ronghui Gu
2020 arXiv   pre-print
G-CLNs extend the Continuous Logic Network (CLN) architecture with gating units and dropout, which allow the model to robustly learn general invariants over large numbers of terms.  ...  In this paper, we introduce a new neural architecture for general SMT learning, the Gated Continuous Logic Network (G-CLN), and apply it to nonlinear loop invariant learning.  ...  This work is sponsored in part by NSF grants CNS  ... 
arXiv:2003.07959v4 fatcat:j3m343yaqjbyhjwadw4swoes4u

Is the cure worse than the disease? overfitting in automated program repair

Edward K. Smith, Earl T. Barr, Claire Le Goues, Yuriy Brun
2015 Proceedings of the 2015 10th Joint Meeting on Foundations of Software Engineering - ESEC/FSE 2015  
In addition to overfitting, we measure the effects of test suite coverage, test suite provenance, and starting program quality, as well as the difference in quality between novice-developer-written and  ...  However, novice developers also overfit, and automated repair performs no worse than these developers.  ...  However, recent work has started to unify the theory underlying G&V repair [61] , suggesting that results from two different techniques may extend to others.  ... 
doi:10.1145/2786805.2786825 dblp:conf/sigsoft/SmithBGB15 fatcat:yr2rj5756jearaona7qcxgbbay

A Generic Framework for Building Dispersion Operators in the Semantic Space [chapter]

Luiz Otavio V. B. Oliveira, Fernando E. B. Otero, Gisele L. Pappa
2018 Genetic Programming Theory and Practice XIII  
An experimental analysis was conducted in a testbed composed of sixteen datasets showing that dispersion operators can improve GSGP search and that the multiplicative version of the operator is overall  ...  better than the additive version.  ...  Acknowledgements The authors would like to thank CAPES, CNPq (141985/2015-1) and Fapemig for their financial support.  ... 
doi:10.1007/978-3-319-97088-2_12 fatcat:uvtybvlfvfglxcz547qpwfk6t4

Theory, Practice, and Applications of Paramagnetic Relaxation Enhancement for the Characterization of Transient Low-Population States of Biological Macromolecules and Their Complexes

G. Marius Clore, Junji Iwahara
2009 Chemical Reviews  
Acknowledgments We thank Attila Szabo and Chun Tang for useful discussions.  ...  Despite its generality, application of the Hwang-Freed theory to macromolecules is extremely complicated and consequently of no practical use.  ...  In other cases, however, PRE effects have been observed to extend well beyond the 15 residue limit and in addition, may not be evenly distributed.  ... 
doi:10.1021/cr900033p pmid:19522502 pmcid:PMC2825090 fatcat:hhhbukpdyfebvgoaaklwqfg3pi

Experimental Aspects of Synthesis

Rüdiger Ehlers
2011 Electronic Proceedings in Theoretical Computer Science  
Furthermore, we discuss which difficulties the complex workflows that begin to appear in modern synthesis tools induce on experimental evaluations and give answers to the question how convincing such evaluations  ...  In particular, we explain why previous tools have incompatible scopes and semantics and provide a framework that reduces the impact of this problem for future experimental comparisons of such tools.  ...  The author wants to thank Barbara Jobstmann and Emmanuel Filiot for helpful comments on the descriptions of the techniques employed in ANZU, LILY and ACACIA.  ... 
doi:10.4204/eptcs.50.1 fatcat:oyu6j7pugjev3eybhzgpujuh4u

S3: syntax- and semantic-guided repair synthesis via programming by examples

Xuan-Bach D. Le, Duc-Hiep Chu, David Lo, Claire Le Goues, Willem Visser
2017 Proceedings of the 2017 11th Joint Meeting on Foundations of Software Engineering - ESEC/FSE 2017  
S3 can successfully and correctly fix at least three times more bugs than the best baseline on datasets of 52 bugs in small programs, and 100 bugs in real-world large programs.  ...  We compare S3's repair effectiveness with state-ofthe-art synthesis engines Angelix, Enumerative, and CVC4.  ...  We treat tests that fail in the before-patched version but pass in the patched version as the failing tests addressed by the bug fixing commit.  ... 
doi:10.1145/3106237.3106309 dblp:conf/sigsoft/LeCLGV17 fatcat:r6t7nhy2bbhhrh6sgkbpw6erqa

Generative Adversarial Networks for Pseudo-Radio-Signal Synthesis [article]

Haythem Chaker, Soumaya Hamouda, Nicola Michailow
2022 arXiv   pre-print
This is simply due to their complexity and degrees of freedom which can be hard to explicitize in compact expressions.  ...  synthesis of subsequently similar 'pseudo-radio-signals'.  ...  output is equivalent to a random noise (underfitting) and to avoid reusing the same pseudo-radio-packet in synthesis (overfitting).  ... 
arXiv:2206.14742v1 fatcat:i2kt6lh75bbr3neue7q5l7ucdq

Type-Directed Synthesis of Products [article]

Jonathan Frankle
2015 arXiv   pre-print
My thesis involved extending this type-directed synthesis engine to handle product types, which required significant modifications to both the underlying theory and the tool itself.  ...  These theoretical enhancements informed a new version of the type-directed synthesis prototype implementation, which remained performance-competitive with the original synthesizer.  ...  In this thesis, I have described extending type-directed synthesis to include product types, which required rethinking both the underlying theory and the prototype implementation.  ... 
arXiv:1510.08121v1 fatcat:keo7iowfvvhfjaegnpf4kdnzum

Sample-level sound synthesis with recurrent neural networks and conceptors

Chris Kiefer
2019 PeerJ Computer Science  
Generation and manipulation of arbitrary patterns using conceptors has significant potential as a sound synthesis method for applications in computer music but has yet to be explored.  ...  The techniques presented here represent an initial exploration of the sound synthesis potential of conceptors, demonstrating possible creative applications in sound design; future possibilities and research  ...  ADDITIONAL INFORMATION AND DECLARATIONS Funding The author received no funding for this work.  ... 
doi:10.7717/peerj-cs.205 pmid:33816858 pmcid:PMC7924416 fatcat:fo4rmbabdjfafapjkeajnpmdpu

Dynamic Bayesian Predictive Synthesis in Time Series Forecasting [article]

Kenichiro McAlinn, Mike West
2017 arXiv   pre-print
We discuss model and forecast combination in time series forecasting.  ...  A foundational Bayesian perspective based on agent opinion analysis theory defines a new framework for density forecast combination, and encompasses several existing forecast pooling methods.  ...  We also thank the editors and four anonymous referees for very detailed, positive and constructive comments on the original version of this paper.  ... 
arXiv:1601.07463v5 fatcat:albd4eqrtzfwbkmtpja53o7jbm

Test-driven synthesis

Daniel Perelman, Sumit Gulwani, Dan Grossman, Peter Provost
2014 SIGPLAN notices  
We present a program synthesizer which can be parameterized by an arbitrary DSL that may contain conditionals and loops and therefore is able to synthesize programs in any domain.  ...  We present applications of our synthesis methodology to end-user programming for transformations over strings, XML, and table layouts.  ...  We thank Nikolai Tillmann and Peli de Halleux of Microsoft Research's Pex4Fun team for access to the Pex4Fun data.  ... 
doi:10.1145/2666356.2594297 fatcat:a4hqx3lxgngs7m2al7myxc3cea
« Previous Showing results 1 — 15 out of 3,189 results