scispace - formally typeset
Search or ask a question
Topic

Class (philosophy)

About: Class (philosophy) is a research topic. Over the lifetime, 821 publications have been published within this topic receiving 28000 citations.


Papers
More filters
Patent
Cyril Goutte1, Eric Gaussier1
09 Feb 2004
TL;DR: In this paper, a method of categorizing objects in which there can be multiple categories of objects and each object can belong to more than one category is described, where each object is assigned to one or more categories in the set.
Abstract: A method of categorizing objects in which there can be multiple categories of objects and each object can belong to more than one category is described. The method defines a set of categories in which at least one category is dependent on another category and then organizes the categories in a hierarchy that embodies any dependencies among them. Each object is assigned to one or more categories in the set. A set of labels corresponding to all combinations of any number of the categories is defined, wherein if an object is relevant to several categories, the object must be assigned the label corresponding to the subset of all relevant categories. Once the new labels are defined, the multi-category, multi-label problem has been reduced to a multi-category, single-label problem, and the categorization task is reduced down to choosing the single best label set for an object.

34 citations

Proceedings ArticleDOI
31 Mar 2022
TL;DR: It is shown that an adversary who can poison a training dataset can cause models trained on this dataset to leak significant private details of training points belonging to other parties, casting doubts on the relevance of cryptographic privacy guarantees in multiparty computation protocols for machine learning, if parties can arbitrarily select their share of training data.
Abstract: We introduce a new class of attacks on machine learning models. We show that an adversary who can poison a training dataset can cause models trained on this dataset to leak significant private details of training points belonging to other parties. Our active inference attacks connect two independent lines of work targeting the integrity and privacy of machine learning training data. Our attacks are effective across membership inference, attribute inference, and data extraction. For example, our targeted attacks can poison <0.1% of the training dataset to boost the performance of inference attacks by 1 to 2 orders of magnitude. Further, an adversary who controls a significant fraction of the training data (e.g., 50%) can launch untargeted attacks that enable 8× more precise inference on all other users' otherwise-private data points. Our results cast doubts on the relevance of cryptographic privacy guarantees in multiparty computation protocols for machine learning, if parties can arbitrarily select their share of training data.

34 citations

Posted Content
TL;DR: In this article, a sharp quantitative version of the $p$-Sobolev inequality with a control on the strongest possible distance from the class of optimal functions was shown.
Abstract: We prove a sharp quantitative version of the $p$-Sobolev inequality for any $12$.

33 citations

Journal ArticleDOI
01 Apr 1991-Mind
TL;DR: In this article, it is argued that explanations of actions in terms of the agent's reasons are essentially teleological explanations, that is, they are explanations which represent the action as an attempt by the agent to achieve some purpose or goal.
Abstract: What amounts to a general account of the class of mental states which have come to be labeled "pro-attitudes" has recently been suggested by Michael Smith (1987). His idea is that we can best understand these as being states "with which the world must fit" and that we can explicate this "direction of fit" metaphor in terms of certain sorts of dispositions of the person who has the state in question. In this paper I want to examine Smith's suggestion. This account of pro-attitudes comes in the context of a defense of the Humean theory of motivation which, according to Smith, can be expressed as follows: "R at t constitutes a motivating reason of agent A to 4 iff there is some v such that R at t consists of a desire of A to P and a belief that were he to 1 he would "' (p. 36). If this theory is to be true it must include a method for ruling out the sort of case where it is claimed that it was a person's mere belief that moved him to do it. Smith tries to do this by arguing for two distinct points. First, he argues that explanations of actions in terms of the agent's reasons are essentially teleological explanations, that is, they are explanations which represent the action as an attempt by the agent to achieve some purpose or goal. So the issue is simply whether the Humean theory is better able to make sense of motivation as pursuit of a goal (p. 44). That this is so, he claims, and this is his second point, will be clear once we understand that only desires, broadly understood, are the right sorts of "mental objects" to make sense of such teleological explanations. The basic idea is that mental states can have one of two possible "directions of fit".' Beliefs are the exemplars of mental states with "mind-to-world" direction of fit since they aim at being true, i.e., "matching the world". If there is a mismatch between a belief and reality then it is the belief which has failed; it has failed to "fit" the world. Desires, on the other side, are exemplars of "world-to-mind" direction of fit. They aim at satisfaction or realization, not truth. If a desire fails to "fit" the world, that is not any defect in the desire. It merely shows that the world needs to be changed if the desire is to be realized. Smith seems correct in holding that "[h]aving a motivating reason is, inter alia, having a goal" (p. 55). The idea of direction of fit of mental states then lets us see, Smith thinks, that having a goal is nothing more than being in a mental state with a world-to-mind direction of fit. "[W]hat kind of state is the having of

33 citations

Book ChapterDOI
01 May 1978
TL;DR: Criteria and methodologies for the conceptual design of databases, particularly in large and sophisticated applications, are addressed and a methodology for designing an abstract syntax specification is outlined.
Abstract: Criteria and methodologies for the conceptual design of databases, particularly in large and sophisticated applications, are addressed. For a design to be understandable to user and designer alike, intuitive methods for abstracting concepts from a mass of detail must be employed. Two abstraction methods are particularly important — aggregation and generalization. Aggregation forms a concept by abstracting a relationship between other concepts (called components). Generalization forms a concept by abstracting a class of other concepts (called categories). The principle of "object relativity" is essential for the successful integration of abstractions. This principle states that individuals, categories, relationships and components are just different ways of viewing the same abstract objects. Using this principle a design may be hierarchically organized into independently meaningful abstractions. An "abstract syntax" is introduced to specify these abstraction hierarchies. An advantage of this abstract syntax is that some concepts do not have to be arbitrarily classified as "roles". The principle of "individual preservation" is a minimal requirement for maintaining the semantics of aggregation and generalization. It states that every user-invokeable operation must preserve the integrity of individuals. A methodology for designing an abstract syntax specification is outlined. The simplicity of this methodology is directly due to the principles of object relativity and individual preservation.

32 citations


Performance
Metrics
No. of papers in the topic in previous years
YearPapers
20241
202311,771
202223,753
2021380
2020186
201962