[신간의 별자리x] 우리/미술, 그리고 ‘슬픔의 박물관’

Despite widespread adoption, machine learning models remain mostly black boxes. Understanding the reasons behind predictions is, however, quite important in assessing trust, which is fundamental if one plans to take action based on a prediction, or when choosing whether to deploy a new model. Such understanding also provides insights into the model, which can be used to transform an untrustworthy model or prediction into a trustworthy one. In this work, we propose LIME, a novel explanation technique that explains the predictions of any classifier in an interpretable and faithful manner, by learning an interpretable model locally varound the prediction. We also propose a method to explain models by presenting representative individual predictions and their explanations in a non-redundant way, framing the task as a submodular optimization problem. We demonstrate the flexibility of these methods by explaining different models for text (e.g. random forests) and image classification (e.g. neural networks). We show the utility of explanations via novel experiments, both simulated and with human subjects, on various scenarios that require trust: deciding if one should trust a prediction, choosing between models, improving an untrustworthy classifier, and identifying why a classifier should not be trusted.

/pdf/why-should-i-trust-you-explaining-the-predictions-of-any-ejtlrtykxr.pdf

"Why Should I Trust You?": Explaining the Predictions of Any Classifier

Deposits of clastic carbonate-dominated (calciclastic) sedimentary slope systems in the rock record have been identified mostly as linearly-consistent carbonate apron deposits, even though most ancient clastic carbonate slope deposits fit the submarine fan systems better. Calciclastic submarine fans are consequently rarely described and are poorly understood. Subsequently, very little is known especially in mud-dominated calciclastic submarine fan systems. Presented in this study are a sedimentological core and petrographic characterisation of samples from eleven boreholes from the Lower Carboniferous of Bowland Basin (Northwest England) that reveals a >250 m thick calciturbidite complex deposited in a calciclastic submarine fan setting. Seven facies are recognised from core and thin section characterisation and are grouped into three carbonate turbidite sequences. They include: 1) Calciturbidites, comprising mostly of highto low-density, wavy-laminated bioclast-rich facies; 2) low-density densite mudstones which are characterised by planar laminated and unlaminated muddominated facies; and 3) Calcidebrites which are muddy or hyper-concentrated debrisflow deposits occurring as poorly-sorted, chaotic, mud-supported floatstones. These

Phd by thesis

Data Mining Practical Machine Learning Tools and Techniques

ing WS1S Systems to Verify Parameterized Networks . . . . . . . . . . . . 188 Kai Baukus, Saddek Bensalem, Yassine Lakhnech and Karsten Stahl FMona: A Tool for Expressing Validation Techniques over Infinite State Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204 J.-P. Bodeveix and M. Filali Transitive Closures of Regular Relations for Verifying Infinite-State Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 Bengt Jonsson and Marcus Nilsson Diagnostic and Test Generation Using Static Analysis to Improve Automatic Test Generation . . . . . . . . . . . . . 235 Marius Bozga, Jean-Claude Fernandez and Lucian Ghirvu Efficient Diagnostic Generation for Boolean Equation Systems . . . . . . . . . . . . 251 Radu Mateescu Efficient Model-Checking Compositional State Space Generation with Partial Order Reductions for Asynchronous Communicating Systems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 Jean-Pierre Krimm and Laurent Mounier Checking for CFFD-Preorder with Tester Processes . . . . . . . . . . . . . . . . . . . . . . . 283 Juhana Helovuo and Antti Valmari Fair Bisimulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 299 Thomas A. Henzinger and Sriram K. Rajamani Integrating Low Level Symmetries into Reachability Analysis . . . . . . . . . . . . . 315 Karsten Schmidt Model-Checking Tools Model Checking Support for the ASM High-Level Language . . . . . . . . . . . . . . 331 Giuseppe Del Castillo and Kirsten Winter Table of

Tools and Algorithms for the Construction and Analysis of Systems. Proc. TACAS 2009

In this paper, ManySAT a new portfolio-based parallel SAT solver is thoroughly described. The design of ManySAT benefits from the main weaknesses of modern SAT solvers: their sensitivity to parameter tuning and their lack of robustness. ManySAT uses a portfolio of complementary sequential algorithms obtained through careful variations of the standard DPLL algorithm. Additionally, each sequential algorithm shares clauses to improve the overall performance of the whole system. This contrasts with most of the parallel SAT solvers generally designed using the divide-and-conquer paradigm. Experiments on many industrial SAT instances, and the first rank obtained by ManySAT in the parallel track of the 2008 SAT-Race clearly show the potential of our design philosophy.

/pdf/manysat-a-parallel-sat-solver-202t38igzn.pdf

ManySAT: a Parallel SAT Solver

Contributors Introduction Lucas Bordeaux, Youssef Hamadi and Pushmeet Kohli Part I. Graphical Structure: 1. Treewidth and hypertree width Georg Gottlob, Gianluigi Greco and Francesco Scarcello 2. Perfect graphs and graphical modeling Tony Jebara Part II. Language Restrictions: 3. Submodular function maximization Andreas Krause and Daniel Golovin 4. Tractable valued constraints Peter G. Jeavons and Stanislav Zivny 5. Tractable knowledge representation formalisms Adnan Darwiche Part III. Algorithms and their Analysis: 6. Tree-reweighted message passing Vladimir Kolmogorov 7. Tractable optimization in machine learning Suvrit Sra 8. Approximation algorithms Mohit Singh and Kunal Talwar 9. Kernelization methods for fixed-parameter tractability Fedor V. Fomin and Saket Saurabh Part IV. Tractability in Some Specific Areas: 10. Efficient submodular function minimization for computer vision Pushmeet Kohli 11. Towards practical graph-based, iteratively decoded channel codes: insights through absorbing sets Lara Dolecek Part V. Heuristics: 12. SAT solvers Joao Marques-Silva and Ines Lynce 13. Tractability and modern satisfiability modulo theories solvers Nikolaj Bjorner and Leonardo de Moura.

Tractability: Practical Approaches to Hard Problems

Machine learning can be used to build models that predict the run-time of search algorithms for hard combinatorial problems. Such empirical hardness models have previously been studied for complete, deterministic search algorithms. In this work, we demonstrate that such models can also make surprisingly accurate predictions of the run-time distributions of incomplete and randomized search methods, such as stochastic local search algorithms. We also show for the first time how information about an algorithm's parameter settings can be incorporated into a model, and how such models can be used to automatically adjust the algorithm's parameters on a per-instance basis in order to optimize its performance. Empirical results for Novelty + and SAPS on structured and unstructured SAT instances show very good predictive performance and significant speedups of our automatically determined parameter settings when compared to the default and best fixed distribution-specific parameter settings.

Performance prediction and automated tuning of randomized and parametric algorithms

Machine learning can be used to build models that predict the run-time of search algorithms for hard combinatorial problems. Such empirical hardness models have previously been studied for complete, deterministic search algorithms. In this work, we demonstrate that such models can also make surprisingly accurate predictions of the run-time distributions of incomplete and randomized search methods, such as stochastic local search algorithms. We also show for the first time how information about an algorithm's parameter settings can be incorporated into a model, and how such models can be used to automatically adjust the algorithm's parameters on a per-instance basis in order to optimize its performance. Empirical results for Novelty+ and SAPS on structured and unstructured SAT instances show very good predictive performance and significant speedups of our automatically determined parameter settings when compared to the default and best fixed distribution-specific parameter settings.

/pdf/performance-prediction-and-automated-tuning-of-randomized-3px5yz2wlp.pdf

Modern programming frameworks come with large libraries, with diverse applications such as for matching regular expressions, parsing XML files and sending email. Programmers often use search engines such as Google and Bing to learn about existing APIs. In this paper, we describe SWIM, a tool which suggests code snippets given API-related natural language queries such as "generate md5 hash code". The query does not need to contain framework-specific trivia such as the type names or methods of interest. We translate user queries into the APIs of interest using clickthrough data from the Bing search engine. Then, based on patterns learned from open-source code repositories, we synthesize idiomatic code describing the use of these APIs. We introduce structured call sequences to capture API-usage patterns. Structured call sequences are a generalized form of method call sequences, with if-branches and while-loops to represent conditional and repeated API usage patterns, and are simple to extract and amenable to synthesis. We evaluated swim with 30 common C# API-related queries received by Bing. For 70% of the queries, the first suggested snippet was a relevant solution, and a relevant solution was present in the top 10 results for all benchmarked queries. The online portion of the workflow is also very responsive, at an average of 1.5 seconds per snippet.

Youssef Hamadi

Papers

ManySAT: a Parallel SAT Solver

Tractability: Practical Approaches to Hard Problems

Performance prediction and automated tuning of randomized and parametric algorithms

Performance prediction and automated tuning of randomized and parametric algorithms

SWIM: synthesizing what i mean: code search and idiomatic snippet synthesis