Pre-trained deep learning models underpin many public-facing applications, and their propensity to reproduce implicit racial and gender stereotypes is an increasing source of concern. The risk of large-scale, unfair outcomes resulting from their use thus raises the need for technical tools to test and audit these systems. In this work, a dataset of 10,000 portrait photographs was generated and classified, using CLIP (Contrastive Language–Image Pretraining), according to six pairs of opposing labels describing a subject’s gender, ethnicity, attractiveness, friendliness, wealth, and intelligence. Label correlation was analyzed and significant associations, corresponding to common implicit stereotypes in culture and society, were found at the 99% significance level. A strong positive correlation was notably found between labels Female and Attractive, Male and Rich, as well as White Person and Attractive. These results are used to highlight the risk of more innocuous labels being used as partial euphemisms for protected attributes. Moreover, some limitations of common definitions of algorithmic fairness as they apply to general-purpose, pre-trained systems are analyzed, and the idea of controlling for bias at the point of deployment of these systems rather than during data collection and training is put forward as a possible circumvention.
This paper characterizes and discusses devolutionary genetic algorithms and evaluates their performances in solving the minimum labeling Steiner tree (MLST) problem. We define devolutionary algorithms as the process of reaching a feasible solution by devolving a population of super-optimal unfeasible solutions over time. We claim that distinguishing them from the widely used evolutionary algorithms is relevant. The most important distinction lies in the fact that in the former type of processes, the value function decreases over successive generation of solutions, thus providing a natural stopping condition for the computation process. We show how classical evolutionary concepts, such as crossing, mutation and fitness can be adapted to aim at reaching an optimal or close-to-optimal solution among the first generations of feasible solutions. We additionally introduce a novel integer linear programming formulation of the MLST problem and a valid constraint used for speeding up the devolutionary process. Finally, we conduct an experiment comparing the performances of devolutionary algorithms to those of state of the art approaches used for solving randomly generated instances of the MLST problem. Results of this experiment support the use of devolutionary algorithms for the MLST problem and their development for other NP-hard combinatorial optimization problems.
Cloud computing is an emerging form of distributed computing that enables a reliable on-demand access to critical Information Technology (IT) services. This flexible access generally allows organizations to improve both their productivity and profitability compared to previous ownership-based computing paradigms. On the other side of the coin, the IT sector as a whole and data centers used in cloud computing infrastructures in particular, are responsible for a significant contribution to global carbon dioxide emissions. Energy consumption and carbon imprints of Cloud infrastructures have thus become a key environmental concern, and several approaches have been proposed with the aim of increasing their energy-efficiency, both within individual data centers and at the level of the infrastructure as a whole. Among these approaches, low-carbon routing algorithms offer operational solutions, which consist in dynamically routing connection requests in order to maximize the use of renewable energy or, equivalently, minimize the use of non- renewable energy. The main claim of this paper is that the existing body of methods suffers from some practical limitations due to the fact that they consider a single objective, at the possible expense of other important aspects such as Quality of Service and cost. Therefore, we propose a bi-criteria routing model, in which both carbon reduction and quality of service, in the form of data transmission delay, are considered as objectives to optimize. Under this model, which to the the best of our knowledge is the fist bi-criteria formulation of the dynamic routing problem in Cloud computing environments, We describe an algorithm to compute efficient supported and non-supported solutions in order to dynamically route connection requests. The application of our algorithm is illustrated through a case study simulated under the same conditions as a previously published work considering a single criterion approach and comparison results are drawn.
Purpose: The purpose of this paper consists in discussing and modeling the asymmetrical nature of evaluation criteria that are notably relevant in strategic planning, project portfolio management and complex systems management. Design/methodology/approach: The author guarantees the theoretic validity of the proposed approach, by showing that it possesses desirable properties such as monotonicity. This allows for a reduced implementation time and thus a real-time use of the model. Findings: The author conducts a comparative study with two other multicriteria decision aid models on a set of real data regarding a power plant sitting problem and highlights some of the limitations of previous models compared to the proposed approach. Originality/value: This paper describes a novel multicriteria decision aid model, entitled SPADE (which is an acronym for Structuring Preferences to Aid DEcisions). It aims at consolidating SWOT analysis with sound decision-theoretic techniques, and making use of its results to support decision-making.
1 January 2015 in International Journal of Applied Management Science
We study the problem of evaluating a set of project proposals and selecting a subset of the best projects among them. Through a review of the relevant literature, we found that the sector in which the analysed projects are carried out impacts the design of the project portfolio selection methodology to be used. In this paper, we focus on the project portfolio evaluation and selection processes in the non-profit sector. We propose new rule-based multi-criteria methodology based on SWOT analysis. The proposed approach is based on an original preference elicitation framework in which a key novelty is the definition of the relative importance of criteria on a two-dimensional scale.