Journal of Computer Languages最新文献

The large-scale adoption of systems that automate classifications using Machine Learning (ML) algorithms raises pressing challenges as they support or make decisions with profound consequences for human beings. It is important to understand how users’ trust is affected by ML models’ suggestions, even when those models are wrong. Many research efforts have focused on the user’s ability to interpret what a model has learned. In this paper, we seek to understand another aspect of ML interpretability: whether and how the presence of classification probabilities and their different distributions are related to users’ trust in model outcomes, especially in ambiguous instances. To this end, we conducted two online surveys in which we asked participants to evaluate their agreement with image classifications of pictures of animals made by an ML model. In the first, we analyze their trust before and after presenting them the model classification probabilities. In the second, we investigate the relationships between class probability distributions and users’ trust in the model. We found that, in some cases, the additional information is correlated with undue trust in the model’s classifications. However, in others, they are associated with inappropriate skepticism.

Access policies specify what are the actions that different actors can perform on available resources. Access policies are a core notion in multiuser environments, such as operating systems and distributed databases. Currently, most of these systems use general data specification languages, such as JSON, XML and YAML to describe access policies. Yet, domain-specific languages are also available for this task. One of such languages is Legalease, from Microsoft. This paper presents a new version of Legalease, called Hapi. Hapi replaces Legalease’s notion of a lattice with a partially ordered set (poset). Posets improve the expressivity of Legalease, at the expenses of a more expensive verification algorithm. This poset-based representation generalizes the notion of actors, actions and resources to user-defined entities. Hapi is publicly available. Its distribution includes a policy visualizer and a code-compression tool to efficiently store specifications.

We propose a novel framework to automatically design and generate exquisite color mandala patterns. First, we summarize pre-requirements for generating reasonable and applicable mandala patterns followed by introducing COOM (Curve, Object, Operation and Model), the fundamental framework of mandala pattern generator. The framework defines how to effectively create visually pleasing mandala sketches. Then we propose a color transfer-based method to colorize the mandala sketch, in which all similar regions will share the same color. Currently, the automatically generated mandala patterns have been used in a mobile game with 3 million DAU (Daily Active Users). The experiments show that our mandala generator is quite effective and comparable with professional artists.

As our lives become increasingly more reliant on software, the impact of its failures grows as well; these failures have diverse causes and their impact ranges from negligible to life-threatening; thus, it is our duty as developers to minimize their occurrence, just as other fields do.

To that end, we build abstractions, move complexity from component to component, and much more, just to stop the end-user from shooting themselves in the foot. However, building said abstractions still requires the original author to know where the pitfalls lie and how to avoid them, an implicit contract that does not constitute a guarantee that they will not shoot themselves and their users in the feet.

Rust aims to minimize the amount of handguns users have at their disposal, locking them behind special unsafe blocks and restricting the set of possible programs through static analysis; this analysis is performed by the compiler which ensures that the program does not contain memory related errors such as use-after-free bugs.

While Rust is able to succeed in the previous domain, other error classes persist, such as errors related to API misusage. Our work aims to tackle that domain, providing a tool which enables developers to write safer APIs using typestates.

We propose a macro which embeds a typestate description DSL in Rust which allows developers to specify typestates for their APIs; the typestate is checked at compile-time for common mistakes and to ensure the correct usage of the typestate, we leverage Rust’s type system.

Our work only requires a Rust compiler, avoiding workflow bloat and keeping the development experience simple; it is open-source and available at https://github.com/rustype/typestate-rs.