The Experts below are selected from a list of 573 Experts worldwide ranked by ideXlab platform

Michael Hanus - One of the best experts on this subject based on the ideXlab platform.

  • putting declarative programming into the web translating curry to javascript
    Principles and Practice of Declarative Programming, 2007
    Co-Authors: Michael Hanus
    Abstract:

    We propose a framework to construct web-oriented user interfaces in a high-level way by exploiting declarative programming techniques. Such user interfaces are intended to manipulate complex data in a type-safe way, i.e., it is ensured that only typecorrect data is accepted by the interface, where types can be specified by standard types of a programming language as well as any Computable Predicate on the data. The interfaces are web based, i.e., the data can be manipulated with standard web browsers without any specific requirements on the client side. However, if the client's browser has JavaScript enabled, one could also check the correctness of the data on the client side providing immediate feedback to the user. In order to release the application programmer from the tedious details to interact with JavaScript, we propose an approach where the programmer must only provide a declarative description of the requirements of the user interface from which the necessary JavaScript programs and HTML forms are automatically generated. This approach leads to a very concise and maintainable implementation of web-based user interfaces. We demonstrate an implementation of this concept in the declarative multi-paradigm language Curry where the integrated functional and logic features are exploited to enable the high level of abstraction proposed in this paper.

  • Putting declarative programming into the web: Translating curry to javascript
    2007
    Co-Authors: Michael Hanus
    Abstract:

    We extend the framework [1] to construct web-oriented user interfaces (WUIs) in a high-level way by exploiting declarative programming techniques. Such user interfaces are intended to manipulate complex data in a type-safe way, i.e., it is ensured that only type-correct data is accepted by the interface, where types can be specified by standard types of a programming language as well as any Computable Predicate on the data (see Fig. 1). The interfaces are web-based, i.e., the data can be manipulated with standard web browsers without any specific requirements on the client side. However, if the client’s browser has JavaScript enabled, one could also check the correctness of the data on the client side providing immediate feedback to the user and reducing network traffic. In order to release the application programmer from the tedious details to interact with JavaScript, we propose an approach where the programmer must only provide a declarative description of the requirements of the user interface from which the necessary JavaScript programs and HTML forms are automatically generated. This approach leads to a very concise and maintainable implementation of web-based user interfaces. We demonstrate an implementation of this concept in the declarative multiparadigm language Curry where the integrated functional and logic features are exploited to enable the high level of abstraction proposed in this paper

  • Type-Oriented Construction of Web User Interfaces
    2006
    Co-Authors: Michael Hanus
    Abstract:

    We propose a new technique for the high-level construction of type-safe web-oriented user interfaces. Our approach is useful to equip applications processing structured data with interfaces to manipulate these data in an efficient and maintainable way. The interfaces are web-based, i.e., the data can be manipulated with standard web browsers without any specific requirements on the client side. In order to support type-safe user interfaces, i.e., interfaces where users can only input type-correct data (types can be standard types of a programming language as well as any Computable Predicate on the data), we propose a set of type-oriented building blocks from which interfaces for more complex types can be easily constructed. This technique leads to a very concise and maintainable implementation of web-based user interfaces.

Translating Curry To Javascript - One of the best experts on this subject based on the ideXlab platform.

  • Interfaces and Presentation]: User Interfaces
    2008
    Co-Authors: Translating Curry To Javascript
    Abstract:

    We propose a framework to construct web-oriented user interfaces in a high-level way by exploiting declarative programming techniques. Such user interfaces are intended to manipulate complex data in a type-safe way, i.e., it is ensured that only type-correct data is accepted by the interface, where types can be specified by standard types of a programming language as well as any Computable Predicate on the data. The interfaces are web-based, i.e., the data can be manipulated with standard web browsers without any specific requirements on the client side. However, if the client’s browser has JavaScript enabled, one could also check the correctness of the data on the client side providing immediate feedback to the user. In order to release the application programmer from the tedious details to interact with JavaScript, we propose an approach where the programmer must only provide a declarative description of the requirements of the user interface from which the necessary JavaScript programs and HTML forms are automatically generated. This approach leads to a very concise and maintainable implementation of web-based user interfaces. We demonstrate an implementation of this concept in the declarative multi-paradigm language Curry where the integrated functional and logic features are exploited to enable the high level of abstraction proposed in this paper

Mahmoody Mohammad - One of the best experts on this subject based on the ideXlab platform.

  • Can Adversarially Robust Learning Leverage Computational Hardness?
    2018
    Co-Authors: Mahloujifar Saeed, Mahmoody Mohammad
    Abstract:

    Making learners robust to adversarial perturbation at test time (i.e., evasion attacks) or training time (i.e., poisoning attacks) has emerged as a challenging task. It is known that for some natural settings, sublinear perturbations in the training phase or the testing phase can drastically decrease the quality of the predictions. These negative results, however, are information theoretic and only prove the existence of such successful adversarial perturbations. A natural question for these settings is whether or not we can make classifiers computationally robust to polynomial-time attacks. In this work, we prove strong barriers against achieving such envisioned computational robustness both for evasion and poisoning attacks. In particular, we show that if the test instances come from a product distribution (e.g., uniform over $\{0,1\}^n$ or $[0,1]^n$, or isotropic $n$-variate Gaussian) and that there is an initial constant error, then there exists a polynomial-time attack that finds adversarial examples of Hamming distance $O(\sqrt n)$. For poisoning attacks, we prove that for any learning algorithm with sample complexity $m$ and any efficiently Computable "Predicate" defining some "bad" property $B$ for the produced hypothesis (e.g., failing on a particular test) that happens with an initial constant probability, there exist polynomial-time online poisoning attacks that tamper with $O (\sqrt m)$ many examples, replace them with other correctly labeled examples, and increases the probability of the bad event $B$ to $\approx 1$. Both of our poisoning and evasion attacks are black-box in how they access their corresponding components of the system (i.e., the hypothesis, the concept, and the learning algorithm) and make no further assumptions about the classifier or the learning algorithm producing the classifier

Mahloujifar Saeed - One of the best experts on this subject based on the ideXlab platform.

  • Can Adversarially Robust Learning Leverage Computational Hardness?
    2018
    Co-Authors: Mahloujifar Saeed, Mahmoody Mohammad
    Abstract:

    Making learners robust to adversarial perturbation at test time (i.e., evasion attacks) or training time (i.e., poisoning attacks) has emerged as a challenging task. It is known that for some natural settings, sublinear perturbations in the training phase or the testing phase can drastically decrease the quality of the predictions. These negative results, however, are information theoretic and only prove the existence of such successful adversarial perturbations. A natural question for these settings is whether or not we can make classifiers computationally robust to polynomial-time attacks. In this work, we prove strong barriers against achieving such envisioned computational robustness both for evasion and poisoning attacks. In particular, we show that if the test instances come from a product distribution (e.g., uniform over $\{0,1\}^n$ or $[0,1]^n$, or isotropic $n$-variate Gaussian) and that there is an initial constant error, then there exists a polynomial-time attack that finds adversarial examples of Hamming distance $O(\sqrt n)$. For poisoning attacks, we prove that for any learning algorithm with sample complexity $m$ and any efficiently Computable "Predicate" defining some "bad" property $B$ for the produced hypothesis (e.g., failing on a particular test) that happens with an initial constant probability, there exist polynomial-time online poisoning attacks that tamper with $O (\sqrt m)$ many examples, replace them with other correctly labeled examples, and increases the probability of the bad event $B$ to $\approx 1$. Both of our poisoning and evasion attacks are black-box in how they access their corresponding components of the system (i.e., the hypothesis, the concept, and the learning algorithm) and make no further assumptions about the classifier or the learning algorithm producing the classifier

Ruppert Eric - One of the best experts on this subject based on the ideXlab platform.

  • Secretive Birds: Privacy in Population Protocols
    2007
    Co-Authors: Delporte-gallet Carole, Fauconnier Hugues, Guerraoui Rachid, Ruppert Eric
    Abstract:

    We study private computations in a system of tiny mobile agents. We consider the mobile population protocol model of Angluin et al. [2] and ask what can be computed without ever revealing any input to a curious adversary. We show that any Computable Predicate of the original population model can be made private through an obfuscation procedure that exploits the inherent non-determinism of the mobility pattern. In short, the idea is for every mobile agent to generate, besides its actual input value, a set of wrong input values to confuse the curious adversary. To converge to the correct result, the procedure has the agents eventually eliminate the wrong values; however, the moment when this happens is hidden from the adversary. This is achieved without jeopardizing the tiny nature of the agents: they still have very small storage size that is independent of the cardinality of the system. We present three variants of this obfuscation procedure that help compute respectively, remainder, threshold, and or Predicates which, when composed, cover all those that can be computed in the population protocol model