Journal articles

1. Arne Decadt, Alexander Erreygers & Jasper De Bock. Conservative decision-making with sets of probabilities: How to infer new choices from previous ones. In Fuzzy Sets and Systems, 109612, Oct. 2025.
   • Abstract
   • paper.pdf
   • doi: https://doi.org/10.1016/j.fss.2025.109612
   We study a gen­er­al­ized ver­sion of max­im­iz­ing ex­pec­ted util­ity, called E-ad­miss­ib­il­ity, to make de­cisions when the de­cision-maker's un­cer­tainty is de­scribed by a set of prob­ab­il­ity mass func­tions. In par­tic­u­lar, in­stead of spe­cify­ing this set dir­ectly, we as­sume that we only have par­tial in­form­a­tion about the de­cision-maker's pref­er­ences or choices, in the form of which op­tions she re­jects from some fi­nite sets of op­tions. We de­scribe both the de­cision-mak­ing pro­cess and the avail­able in­form­a­tion us­ing choice func­tions, and we provide an al­gorithm, based on lin­ear pro­gram­ming, to com­pute the most con­ser­vat­ive ex­ten­sion of a giv­en choice as­sess­ment to a choice func­tion that makes de­cisions based on E-ad­miss­ib­il­ity. Next, we re­late this E-ad­miss­ible ex­ten­sion to the so-called co­her­ent ex­ten­sion and show how the same tech­niques that are used to sim­pli­fy the com­pu­ta­tion of this co­her­ent ex­ten­sion can also be used to sim­pli­fy that of the E-ad­miss­ible one. In our ex­per­i­ments, we demon­strate that de­cision-mak­ing with the E-ad­miss­ible ex­ten­sion is faster and more in­form­at­ive than with the co­her­ent one, but also ob­serve that the re­quired com­pu­ta­tions are chal­len­ging once the para­met­ers of the prob­lem scale.
2. Arne Decadt, Alexander Erreygers & Jasper De Bock. Extending choice assessments to choice functions: An algorithm for computing the natural extension. In International Journal of Approximate Reasoning, 178:109331, Mar. 2025.
   • Abstract
   • paper.pdf
   • doi: https://doi.org/10.1016/j.ijar.2024.109331
   • arXiv: 2407.21164
   We study how to in­fer new choices from pri­or choices us­ing the frame­work of choice func­tions, a uni­fy­ing math­em­at­ic­al frame­work for de­cision-mak­ing based on sets of pref­er­ence or­ders. In par­tic­u­lar, we define the nat­ur­al (most con­ser­vat­ive) ex­ten­sion of a giv­en choice as­sess­ment to a co­her­ent choice func­tion—whenev­er pos­sible—and use this nat­ur­al ex­ten­sion to make new choices. We provide a prac­tic­al al­gorithm for com­put­ing this nat­ur­al ex­ten­sion and vari­ous ways to im­prove scalab­il­ity. Fi­nally, we test these al­gorithms for different types of choice as­sess­ments.