Filters








5,315 Hits in 7.3 sec

Approximate and Probabilistic Computing: Design, Coding, Verification (Dagstuhl Seminar 15491)

Antonio Filieri, Marta Kwiatkowska, Sasa Misailovic, Todd Mytkowicz, Marc Herbstritt
2016 Dagstuhl Reports  
The aim of this seminar was to bring together academic and industrial researchers from the areas of probabilistic model checking, quantitative software analysis, probabilistic programming, and approximate  ...  Millions of people already use software which computes with and reasons about approximate/probabilistic data daily.  ...  Arguably most work on the problem of program synthesis is based on various models based in discrete structures, e.g. related to model checking, game theoretic models, combinatorial optimisation, etc.  ... 
doi:10.4230/dagrep.5.11.151 dblp:journals/dagstuhl-reports/FilieriKMM15 fatcat:dao63covdjhflma6fflt3meus4

A Survey of Machine Learning for Big Code and Naturalness

Miltiadis Allamanis, Earl T. Barr, Premkumar Devanbu, Charles Sutton
2018 ACM Computing Surveys  
We present a taxonomy based on the underlying design principles of each model and use it to navigate the literature.  ...  Research at the intersection of machine learning, programming languages, and software engineering has recently taken important steps in proposing learnable probabilistic models of source code that exploit  ...  [116] use a graphical model to learn commonalities of programs across similar tasks with the aim to improve program synthesis search. Menon et al.  ... 
doi:10.1145/3212695 fatcat:iuuocyctg5adjmobhc2zw23rfu

A Survey of Machine Learning for Big Code and Naturalness [article]

Miltiadis Allamanis, Earl T. Barr, Premkumar Devanbu, Charles Sutton
2018 arXiv   pre-print
We present a taxonomy based on the underlying design principles of each model and use it to navigate the literature.  ...  Research at the intersection of machine learning, programming languages, and software engineering has recently taken important steps in proposing learnable probabilistic models of source code that exploit  ...  [104] use a graphical model to learn commonalities of programs across similar tasks with the aim to improve program synthesis search. Menon et al.  ... 
arXiv:1709.06182v2 fatcat:hbvgyonqsjgq3nqwji6jf3aybe

Active Learning through Adaptive Heterogeneous Ensembling

Zhenyu Lu, Xindong Wu, Josh C. Bongard
2015 IEEE Transactions on Knowledge and Data Engineering  
C. (2012) Learning Comparative User Models for Accelerating HumanComputer Collaborative Search.  ...  C. (2013) Improving Genetic Programming Based Symbolic Regression Using Deterministic Machine Learning. Procs of the IEEE Congress on Evolutionary Computation, Cancun, MX. 49.  ... 
doi:10.1109/tkde.2014.2304474 fatcat:zgibjwqwengqrioqvn5e2t4zqm

Scaling Neural Program Synthesis with Distribution-based Search [article]

Nathanaël Fijalkow and Guillaume Lagarde and Théo Matricon and Kevin Ellis and Pierre Ohlmann and Akarsh Potta
2021 arXiv   pre-print
We investigate how to augment probabilistic and neural program synthesis methods with new search algorithms, proposing a framework called distribution-based search.  ...  Collectively these findings offer theoretical and applied studies of search algorithms for program synthesis that integrate with recent developments in machine-learned program synthesizers.  ...  Distribution-based search We work within the syntax guided program synthesis (SyGuS) framework introduced by Alur et al. (2013) , see also .  ... 
arXiv:2110.12485v1 fatcat:2a6rudsktraohmeni7ehsljs5i

Towards Verified Artificial Intelligence [article]

Sanjit A. Seshia, Dorsa Sadigh, S. Shankar Sastry
2020 arXiv   pre-print
Verified artificial intelligence (AI) is the goal of designing AI-based systems that that have strong, ideally provable, assurances of correctness with respect to mathematically-specified requirements.  ...  (FCRP) a Semiconductor Research Corporation program sponsored by MARCO and DARPA, by the DARPA BRASS and Assured Autonomy programs, by Toyota under the iCyPhy center, and by Berkeley Deep Drive.  ...  CNS-1545126 (VeHICaL), CNS-1646208, and CCF-1837132 (FMitF), by an NDSEG Fellowship, by the TerraSwarm Research Center, one of six centers supported by the STARnet phase of the Focus Center Research Program  ... 
arXiv:1606.08514v4 fatcat:ozoldsdnzjghddhwz5xju6zqvu

Safe Autonomy Under Perception Uncertainty Using Chance-Constrained Temporal Logic

Susmit Jha, Vasumathi Raman, Dorsa Sadigh, Sanjit A. Seshia
2017 Journal of automated reasoning  
These systems often operate in uncertain environments and in the presence of noisy sensors, and use machine learning and statistical sensor fusion algorithms to form an internal model of the world that  ...  We present a novel automated synthesis technique that compiles C2TL specification into mixed integer constraints, and uses second-order (quadratic) cone programming to synthesize optimal control of autonomous  ...  Where applicable, we use a baseline comprised of LQG-based motion planning algorithm [46] and a Monte Carlo sampling-based search algorithm to find an optimal trajectory over the uncertain world model  ... 
doi:10.1007/s10817-017-9413-9 fatcat:wfb2bqprcbhd5k6iapjb3yiddm

On Repair with Probabilistic Attribute Grammars [article]

Manos Koukoutos, Mukund Raghothaman, Etienne Kneuss, Viktor Kuncak
2017 arXiv   pre-print
Among most promising ideas emerging for synthesis are syntax-driven search, probabilistic models of code, and the use of input-output examples.  ...  We show that synthesis in this framework can be viewed as an instance of graph search, permitting the use of well-understood families of techniques such as A*.  ...  Our paper is an attempt to apply similar techniques to accelerate program synthesis.  ... 
arXiv:1707.04148v1 fatcat:goxtgzjq4raolhrcwzkpxchvgm

M3: Semantic API Migrations [article]

Bruce Collie, Philip Ginsbach, Jackson Woodruff, Ajitha Rajan, Michael O'Boyle
2020 arXiv   pre-print
To tackle this problem, this paper proposes a novel approach (M^3), where probabilistic program synthesis is used to semantically model the behavior of library functions.  ...  Then, we use an SMT-based code search engine to discover similar code in user applications. These discovered instances provide potential locations for API migrations.  ...  Model uses component-based sketching [24] together with novel learned probabilistic models to efficiently search for the most likely structure for correct solutions.  ... 
arXiv:2008.12118v1 fatcat:wqzks3q2bfhgngul3jeymisjbq

Omega: An Architecture for AI Unification [article]

Eray Özkural
2018 arXiv   pre-print
We retain the basic design of a fundamental algorithmic substrate called an "AI kernel" for problem solving and basic cognitive functions like memory, and a larger, modular architecture that re-uses the  ...  At the highest scale, the system runs the most expensive model-free learning algorithms that can search over architectures, models, and components, and updates its persistent, long term memory based on  ...  It must have practically effective training methods for learning tasks, such as the GPU accelerated training methods used in deep learning.  ... 
arXiv:1805.12069v1 fatcat:gkj27nzfb5azrkfrawgqil5pim

A tensorized logic programming language for large-scale data [article]

Ryosuke Kojima, Taisuke Sato
2019 arXiv   pre-print
We introduce a new logic programming language T-PRISM based on tensor embeddings.  ...  Combing these two parts provides a remarkably wide range of high-level declarative modeling from symbolic reasoning to deep learning.  ...  We explained how programs are compiled into tensor equations by way of explanation graphs with Einstein's notation using Prolog's tabled search, and executed on TensorFlow efficiently.  ... 
arXiv:1901.08548v1 fatcat:bkl6kr43gng5zpgsc4bdisftvi

A Low-Cost Robot Science Kit for Education with Symbolic Regression for Hypothesis Discovery and Validation [article]

Logan Saar, Haotong Liang, Alex Wang, Austin McDannald, Efrain Rodriguez, Ichiro Takeuchi, A. Gilad Kusne
2022 arXiv   pre-print
To build and use these systems, the next generation workforce requires expertise in diverse areas including ML, control systems, measurement science, materials synthesis, decision theory, among others.  ...  We discuss its use in the course and its greater capability to teach the dual tasks of autonomous model exploration, optimization, and determination, with an example of autonomous experimental "discovery  ...  Active learning-based recommendation engines guide experiments both in the lab and in silico, accelerating the discovery of novel materials.  ... 
arXiv:2204.04187v2 fatcat:deassb63dzfwfkig5xx4chykse

Neurally-Guided Structure Inference [article]

Sidi Lu and Jiayuan Mao and Joshua B. Tenenbaum and Jiajun Wu
2019 arXiv   pre-print
We evaluate our algorithm on two representative structure inference tasks: probabilistic matrix decomposition and symbolic program parsing.  ...  It outperforms data-driven and search-based alternatives on both tasks.  ...  Guided search. Data-driven models such as neural networks can be used to improve the efficiency of search-based inference.  ... 
arXiv:1906.07304v2 fatcat:7ig4zqki6rgipc6vrgoqyph7ay

Model-Driven Synthesis for Programming Tutors [article]

Niek Mulleners, Johan Jeuring, Bastiaan Heeren
2020 arXiv   pre-print
We propose to investigate how we can overcome this problem by using program synthesis, which we use to generate correct solutions that closely match a student program, and give feedback based on the results  ...  On the other hand, we want a student to write her program in such a way that we can provide constructive feedback.  ...  [10] accelerate search-based synthesis by performing A* search [7] , with a heuristic based on learned probabilistic models.  ... 
arXiv:2011.07510v1 fatcat:g2tbcw5qtjctvgsc22nxdmrm3y

Programmatically Interpretable Reinforcement Learning [article]

Abhinav Verma, Vijayaraghavan Murali, Rishabh Singh, Pushmeet Kohli, Swarat Chaudhuri
2019 arXiv   pre-print
Unlike the popular Deep Reinforcement Learning (DRL) paradigm, which represents policies by neural networks, PIRL represents policies using a high-level, domain-specific programming language.  ...  NDPS works by first learning a neural policy network using DRL, and then performing a local search over programmatic policies that seeks to minimize a distance from this neural "oracle".  ...  However, unlike SYGUS and previous sketch-based synthesis approaches that use logical constraints as specification, PIRL searches for policies with quantitative objectives. Imitation Learning.  ... 
arXiv:1804.02477v3 fatcat:3hydz34zurb4xcr35v6hlkmuvi
« Previous Showing results 1 — 15 out of 5,315 results