Constructing a Fair Classifier with Generated Fair Data
نویسندگان
چکیده
Fairness in machine learning is getting rising attention as it directly related to real-world applications and social problems. Recent methods have been explored alleviate the discrimination between certain demographic groups that are characterized by sensitive attributes (such race, age, or gender). Some studies found data itself biased, so training on causes unfair decision making. Models trained raw can replicate even exacerbate bias prediction groups. This leads vastly different performance In order address this issue, we propose a new approach improve fairness generating fair data. We introduce generative model generate cross-domain samples w.r.t. multiple attributes. ensures infinite number of balanced \wrt both target label enhance prediction. By classifier solely with synthetic then transfer real data, overcome under-representation problem which non-trivial since collecting extremely time resource consuming. provide empirical evidence demonstrate benefit our respect accuracy.
منابع مشابه
Constructing Fair-Exchange P2P File Market
P2P is a promising technology to construct the underlying supporting layer of a Grid. It is known that contribution from all the peers is vital for the sustainability of a P2P community, but peers are often selfish and unwilling to contribute. In this paper we describe how to construct a fair file-exchanging P2P community. We name this community a P2P file market. Our scheme forces peers to con...
متن کاملAn intelligent web-page classifier with fair feature-subset selection
The explosion of on-line information has given rise to many manually constructed topic hierarchies (such as Yahoo!!). But with the current growth rate in the amount of information, manual classification in topic hierarchies results in an immense information bottleneck. Therefore, developing an automatic classifier is an urgent need. However, the classifiers suffer fkom the enormous dimensionali...
متن کاملFAIR: Fair Audience InfeRence
Given the recent changes in the policy governing Internet content distribution, such as the institution of per listener royalties for Internet radio broadcasters, content distributors now have an incentive to under-report the size of their audience. Previous audience measurement schemes only protect against inflation of audience size. We present the first protocols for audience measurement that...
متن کاملFair Is Not Fair Everywhere.
Distributing the spoils of a joint enterprise on the basis of work contribution or relative productivity seems natural to the modern Western mind. But such notions of merit-based distributive justice may be culturally constructed norms that vary with the social and economic structure of a group. In the present research, we showed that children from three different cultures have very different i...
متن کاملHow Fair is Fair Trade?
This paper investigates to what extent fair trade programmes, are indeed ‘fair’. This is accomplished by comparing fair trade with free trade and protectionist trade regimes on their compliance of the criteria set by the fair trade movement itself. This comparison is made using comparative cost based and economies of scale models. It is found that whether or not fair trade is superior to free t...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: Proceedings of the ... AAAI Conference on Artificial Intelligence
سال: 2021
ISSN: ['2159-5399', '2374-3468']
DOI: https://doi.org/10.1609/aaai.v35i9.16965