EURO Journal on Decision Processes最新文献

Decision-making is crucial to the performance and well-being of any organization. While artificial intelligence algorithms are increasingly used in the industry for decision-making purposes, the adoption of decision-making techniques to develop new artificial intelligence models does not follow the same trend. Complex artificial intelligence algorithm structures such as gradient boosting, ensembles, and neural networks offer higher accuracy at the expense of transparency. In organizations, however, managers and other stakeholders need to understand how an algorithm came to a given decision to properly criticize, learn from, audit, and improve said algorithms. Among the most recent techniques to address this, explainable artificial intelligence (XAI) algorithms offer a previously unforeseen level of interpretability, explainability, and informativeness to different human roles in the industry. XAI algorithms seek to balance the trade-off between interpretability and accuracy by introducing techniques that, for instance, explain the feature relevance in complex algorithms, generate counterfactual examples in “what-if?” analyses, and train surrogate models that are intrinsically explainable. However, while the trade-off between these two objectives is commonly referred to in the literature, only some proposals use multi-objective optimization in XAI applications. Therefore, this document proposes a new multi-objective optimization application to help decision-makers (for instance, data scientists) to generate new surrogate machine learning models based on black-box models. These surrogates are generated by a multi-objective problem that maximizes, at the same time, interpretability and accuracy. The proposed application also has a multi-criteria decision-making step to rank the best surrogates considering these two objectives. Results from five classification and regression datasets tested on four black-box models show that the proposed method can create simple surrogates maintaining high levels of accuracy.

Modern software relies heavily on data and machine learning, and affects decisions that shape our world. Unfortunately, recent studies have shown that because of biases in data, software systems frequently inject bias into their decisions, from producing more errors when transcribing women’s than men’s voices to overcharging people of color for financial loans. To address bias in software, data scientists and software engineers need tools that help them understand the trade-offs between model quality and fairness in their specific data domains. Toward that end, we present fairkit-learn, an interactive toolkit for helping engineers reason about and understand fairness. Fairkit-learn supports over 70 definition of fairness and works with state-of-the-art machine learning tools, using the same interfaces to ease adoption. It can evaluate thousands of models produced by multiple machine learning algorithms, hyperparameters, and data permutations, and compute and visualize a small Pareto-optimal set of models that describe the optimal trade-offs between fairness and quality. Engineers can then iterate, improving their models and evaluating them using fairkit-learn. We evaluate fairkit-learn via a user study with 54 students, showing that students using fairkit-learn produce models that provide a better balance between fairness and quality than students using scikit-learn and IBM AI Fairness 360 toolkits. With fairkit-learn, users can select models that are up to 67% more fair and 10% more accurate than the models they are likely to train with scikit-learn.

Modern discussions of multiple criteria decision-making extend back about half a century. I reflect on key developments, schools of thought and controversies that have taken place over the period, arguing that perhaps those of us in different schools focus too much on our differences and do not capitalise enough on what we share in common. Moreover, the differences between schools are indications of their respective weaknesses and can drive improvements in each. The discussion points to a number of issues and research needs that the community needs to address.

Automated decision systems are increasingly used to take consequential decisions in problems such as job hiring and loan granting with the hope of replacing subjective human decisions with objective machine learning (ML) algorithms. However, ML-based decision systems are prone to bias, which results in yet unfair decisions. Several notions of fairness have been defined in the literature to capture the different subtleties of this ethical and social concept (e.g., statistical parity, equal opportunity, etc.). Fairness requirements to be satisfied while learning models created several types of tensions among the different notions of fairness and other desirable properties such as privacy and classification accuracy. This paper surveys the commonly used fairness notions and discusses the tensions among them with privacy and accuracy. Different methods to address the fairness-accuracy trade-off (classified into four approaches, namely, pre-processing, in-processing, post-processing, and hybrid) are reviewed. The survey is consolidated with experimental analysis carried out on fairness benchmark datasets to illustrate the relationship between fairness measures and accuracy in real-world scenarios.

In this paper, a bi-objective mathematical model is presented for the problem of designing a resilient supply chain network of pharmaceutical-health relief items in the condition of disruption, taking into account the possibility of lateral transshipment. The first objective function of the model aims to minimize the total costs and considering the importance of effective and efficient distribution to meet the demand of patients in a humanitarian supply chain network, the second objective function is minimizing the total time required to deliver relief items to the demand points. Given the inherent uncertainty associated with the occurrence of a crisis and its impact on the supply chain network, a scenario-based robust optimization method is used to address the problem. The model is solved using the epsilon constraint method for small sizes and the NSGA-II meta-heuristic algorithm for larger sizes. In addition, the model is solved with and without lateral transshipment, and the results are compared and analyzed. The findings indicate that lateral transshipment can improve the performance of the supply chain and reduce the level of shortage.

In this paper we examine a multi-period portfolio optimization problem in a fuzzy environment. The proposed optimization model is subject to CVaR constraint, transaction constraint and cardinality constraint. The returns of the assets are assumed to be trapezoidal fuzzy variables and therefore the portfolio return and risk are quantified by the possibilistic mean and semivariance of the fuzzy returns respectively. A dynamic programming method is used to solve the proposed mixed interger optimization model for different cardinality constraints. A numerical study based on real stocks market data is provided to test the efficiency of the proposed algorithm. The sensitivity of the optimal portfolio investment strategies is tested for different confidence levels for the CVaR constraint.

The paper offers a contribution to the interdisciplinary constructs of analyzing fairness issues in automatic algorithmic decisions. Section 2 shows that technical choices in supervised learning have social implications that need to be considered. Section 3 proposes a contextual approach to the issue of unintended group discrimination, i.e. decision rules that are facially neutral but generate disproportionate impacts across social groups (e.g., gender, race or ethnicity). The contextualization will focus on the legal systems of the United States on the one hand and Europe on the other. In particular, legislation and case law tend to promote different standards of fairness on both sides of the Atlantic. Section 4 is devoted to the explainability of algorithmic decisions; it will confront and attempt to cross-reference legal concepts (in European and French law) with technical concepts and will highlight the plurality, even polysemy, of European and French legal texts relating to the explicability of algorithmic decisions. The conclusion proposes directions for further research.