Search

This paper studies a parametric family of algorithmic fairness metrics, called generalized entropy, which originally has been used in public welfare and recently introduced to machine learning community. As a meaningful metric to evaluate algorithmic fairness, it requires that generalized entropy specify fairness requirements of a classification problem and the fairness requirements should be realized with small deviation by an algorithm. We investigate the role of generalized entropy as a design parameter for fair classification algorithm through a fair empirical risk minimization with a constraint specified in terms of generalized entropy. We theoretically and experimentally study learnability of the problem., Comment: 56pages and 92 figures Revised for adding experimental results

Cellular networks are facing severe traffic overloads due to the proliferation of smart handheld devices and traffic-hungry applications. A cost-effective and practical solution is to offload cellular data through WiFi. Recent theoretical and experimental studies show that a scheme, referred to as delayed WiFi offloading, can significantly save the cellular capacity by delaying users' data and exploiting mobility and thus increasing chance of meeting WiFi APs (Access Points). Despite a huge potential of WiFi offloading in alleviating mobile data explosion, its success largely depends on the economic incentives provided to users and operators to deploy and use delayed offloading. In this paper, we study how much economic benefits can be generated due to delayed WiFi offloading, by modeling a market based on a two-stage sequential game between a monopoly provider and users. We also provide extensive numerical results computed using a set of parameters from the real traces and Cisco's projection of traffic statistics in year 2015. In both analytical and numerical results, we model a variety of practical scenarios and control knobs in terms of traffic demand and willingness to pay of users, spatio-temporal dependence of pricing and traffic, and diverse pricing and delay tolerance. We demonstrate that delayed WiFi offloading has considerable economic benefits, where the increase ranges from 21% to 152% in the provider's revenue, and from 73% to 319% in the users' surplus, compared to on-the-spot WiFi offloading., Comment: 15 pages, 11 figures, submitted for publication

In this paper, we consider medium access control of local area networks (LANs) under limited-information conditions as befits a distributed system. Rather than assuming "by rule" conformance to a protocol designed to regulate packet-flow rates (e.g., CSMA windowing), we begin with a non-cooperative game framework and build a dynamic altruism term into the net utility. The effects of altruism are analyzed at Nash equilibrium for both the ALOHA and CSMA frameworks in the quasistationary (fictitious play) regime. We consider either power or throughput based costs of networking, and the cases of identical or heterogeneous (independent) users/players. In a numerical study we consider diverse players, and we see that the effects of altruism for similar players can be beneficial in the presence of significant congestion, but excessive altruism may lead to underuse of the channel when demand is low.

We consider a slotted-ALOHA LAN with loss-averse, noncooperative greedy users. To avoid non-Pareto equilibria, particularly deadlock, we assume probabilistic loss-averse behavior. This behavior is modeled as a modulated white noise term, in addition to the greedy term, creating a diffusion process modeling the game. We observe that when player's modulate with their throughput, a more efficient exploration of play-space results, and so finding a Pareto equilibrium is more likely over a given interval of time.