1,721,057 research outputs found
A logic of negative trust
Full text linkWe present a logic to model the behaviour of an agent trusting or not trusting messages sent by another agent. The logic formalises trust as a consistency checking function with respect to currently available information. Negative trust is modelled in two forms: distrust, as the rejection of incoming inconsistent information; mistrust, as revision of previously held information becoming undesirable in view of new incoming inconsistent information, which the agent wishes to accept. We provide a natural deduction calculus, a relational semantics and prove soundness and completeness results. We overview a number of applications which have been investigated for the proof-theoretical formulation of the logic
Proof-checking Bias in Labeling Methods
Full text linkWe introduce a typed natural deduction system designed to formally verify the presence of bias in automatic labeling methods. The system relies on a ”data-as-terms” and ”labels-as-types” interpretation of formulae, with derivability contexts encoding probability distributions on training data. Bias is understood as the divergence that expected probabilistic labeling by a classifier trained on opaque data displays from the fairness constraints set by a transparent dataset
HTLC: Hyperintensional Typed Lambda Calculus
Full text linkIn this paper we introduce the logic HTLC, for Hyperintensional Typed Lambda Calculus. The system extends the typed lambda-calculus with hyperintensions and related rules. The polymorphic nature of the system allows to reason with expressions for extensional, intensional and hyperintentsional entities. We inspect meta-theoretical properties and show that HTLC is complete in Henkin's sense under a weakening of the cardinality constraint for the domain of hyperintensions
Miscomputation
No full textThe phenomenon of digital computation is explained (often differently) in computer science, computer engineering and more broadly in cognitive science. Although the semantics and implications of malfunctions have received attention in the philosophy of biology and philosophy of technology, errors in computational systems remain of interest only to computer science. Miscomputation has not gotten the philosophical attention it deserves. Our paper fills this gap by offering a taxonomy of miscomputations. This taxonomy is underpinned by a conceptual analysis of the design and implementation of conventional computational systems at various levels of abstraction. It shows that 'malfunction' as it is typically used in the philosophy of artefacts only represents one type of miscomputation
Prior to Trust: Frequentist and Bayesian views of Trust in AI
Full text linkThe notions of trust and trustworthiness in the field of AI are currently the focus of a collective, interdisciplinary effort for clarification. In this work, we contribute to this ongoing debate by identifying two senses in which an agent might place trust in an AI system. The first sense, referring to trustworthiness as formalised in previous work, considers the results of tests conducted on the system alongside the agent’s expectations. The second sense, extends the former by factoring in the agent’s “pragmatic” background when considering these tests. We argue that these two forms of trust can be understood in relation to well-known approaches in statistical inference: the first aligns with a frequentist interpretation, while the second reflects a Bayesian view of trust
Data Quality Dimensions for Fair AI
Full text linkArtificial Intelligence (AI) systems are not intrinsically neutral and biases trickle in any type of tech-
nological tool. In particular when dealing with people, the impact of AI algorithms’ technical errors
originating with mislabeled data is undeniable. As they feed wrong and discriminatory classifications,
these systems are not systematically guarded against bias. In this article we consider the problem of
bias in AI systems from the point of view of data quality dimensions. We highlight the limited model
construction of bias mitigation tools based on accuracy strategy, illustrating potential improvements of a
specific tool in gender classification errors occurring in two typically difficult contexts: the classification
of non-binary individuals, for which the label set becomes incomplete with respect to the dataset; and
the classification of transgender individuals, for which the dataset becomes inconsistent with respect
to the label set. Using formal methods for reasoning about the behavior of the classification system in
presence of a changing world, we propose to reconsider the fairness of the classification task in terms of
completeness, consistency, timeliness and reliability, and offer some theoretical results
The Philosophy of Computer Science, in The Stanford Encyclopedia of Philosophy (Spring 2021 Edition), Edward N. Zalta (ed.)
No full textThe philosophy of computer science is concerned with the ontological and methodological issues arising from within the academic discipline of computer science, and from the practice of software development and its commercial and industrial deployment. More specifically, the philosophy of computer science considers the ontology and epistemology of computational systems, focusing on problems associated with their specification, programming, implementation, verification and testing. The complex nature of computer programs ensures that many of the conceptual questions raised by the philosophy of computer science have related ones in the philosophy of mathematics, the philosophy of empirical sciences, and the philosophy of technology. We shall provide an analysis of such topics that reflects the layered nature of the ontology of computational systems in Sections 1–5; we then discuss topics involved in their methodology in Sections 6–8
Epistemic Modalities
Full text linkI present an analysis of the notion of epistemic modalities, based on an appropriate interpretation of two basic constructivist issues: verification and epistemic agency. Starting from an historical analysis of conditions for judgments, I analyze first the reading of necessity with respect to apodictic judgements, and then that of possibility with respect to hypothetical judgement. The analysis results in a formal treatment of rules for judgemental modal operators, whose aim is to preserve epistemic states corresponding to verified and unverified assumptions in contexts. In the conclusion, further tracks of research are indicated for designing a semantic framework and defining multi-agents systems
A typed natural deduction calculus to reason about secure trust
Full text linkSystem integrity can be put at risk by unintentional transitivity of resource access. We present a natural deduction calculus for an access control model with an explicit trust function on resources. Its inference relation is designed to limit unintentionally transitive access from untrusted parties. We also offer results for ordered cut and normalization related to security and hint at a prototype implementation
- …
