A flexible class of dependence-aware multi-label loss functions

The idea to exploit label dependencies for better prediction is at the core of methods for multi-label classification (MLC), and performance improvements are normally explained in this way. Surprisingly, however, there is no established methodology that allows to analyze the dependence-awareness of MLC algorithms. With that goal in mind, we introduce a class of loss functions that are able to capture the important aspect of label dependence. To this end, we leverage the mathematical framework of non-additive measures and integrals. Roughly speaking, a non-additive measure allows for modeling the importance of correct predictions of label subsets (instead of single labels), and thereby their impact on the overall evaluation, in a flexible way. The well-known Hamming and subset 0/1 losses are rather extreme special cases of this function class, which give full importance to single label sets or the entire label set, respectively. We present concrete instantiations of this class, which appear to be especially appealing from a modeling perspective. The assessment of multi-label classifiers in terms of these losses is illustrated in an empirical study, clearly showing their aptness at capturing label dependencies. Finally, while not being the main goal of this study, we also show some preliminary results on the minimization of this parametrized family of losses.

Symplectic Geometry of the Koopman Operator

We consider the Koopman operator generated by an invertible transformation of a space with a finite countably additive measure. If the square of this transformation is ergodic, then the orthogonal Koopman operator is a symplectic transformation on the real Hilbert space of square summable functions with zero mean. An infinite set of quadratic invariants of the Koopman operator is specified, which are pairwise in involution with respect to the corresponding symplectic structure. For transformations with a discrete spectrum and a Lebesgue spectrum, these quadratic invariants are functionally independent and form a complete involutive set, which suggests that the Koopman transform is completely integrable.

Wirtinger integral inequalities for pseudo-integrals and pseudo-additive measure

The main purpose of this paper is to show Wirtinger type inequalities for the pseudo-integral. We are concerned with pseudo-integrals based on the following three canonical cases: in the first case, the real semiring with pseudo-operation is generated by a strictly monotone continuous function g; in the second case, the pseudo-operations include a pseudo-multiplication and a power arithmetic addition; in the last case, ⊕-measures are interval-valued. Examples are given to illustrate these equalities.

Structurally Stable Symmetric Tilings on the Plane

A symmetric m-tilings model on the plane is assembled to be a phase portrait for a structurally stable Hamiltonian system. Integral of the system is the quasi-periodic function with m-fold rotational symmetry being result of the semi-dynamic system action on the unit interval. Some examples for pentagonal and heptagonal tilings has been built in detail. Some properties of an additive measure and order for tilings have been discussed.

The Levelling-Down Objection and the additive measure of the badness of inequality

The Levelling-Down Objection is a standard objection to monistic egalitarian theories where equality is the only thing that has intrinsic value. Most egalitarians, however, are value pluralists; they hold that, in addition to equality being intrinsically valuable, the egalitarian currency in which we are equal or unequal is also intrinsically valuable. In this paper, I argue that the Levelling-Down Objection still minimizes the weight that the intrinsic badness of inequality could have in the overall intrinsic evaluation of outcomes, given a certain way of measuring the badness of inequality, namely, the Additive Individual-Complaints Measure.

A Novel Algorithm Based on 2-Additive Measure and Shapley Value and Its Application in Land Pollution Remediation

In this article, a new aggregation operator called the Young–Shapley optimal weight (Y-SOW) operator is proposed to aggregate heterogeneous information in group decision-making. The Y-SOW operator combines the Shapley value with the Young inequality. Meanwhile, a series of special cases and main properties of the Y-SOW operator are studied. Furthermore, the dispersion maximization model of the Y-SOW operator is established to obtain the optimal 2-additive measure. In the Shapley value method of the cooperative game, the 2-additive measure not only simplifies the complexity of fuzzy measures but also solves the interaction between attributes. The Shapley value of the 2-additive measure is explored to the weight of the Y-SOW operator. Finally, the Y-SOW operator-based multiattribute group decision (YSMGAD) algorithm is proposed. The application of the YSMGAD algorithm for land pollution remediation is analyzed.

New Composite Measure for ADL Limitations: Application to Predicting Nursing Home Placement for Michigan MI Choice Clients

Functional status measured by activities of daily living (ADL) may be used to predict nursing home placement. Scoring of ADL measures is summarized for convenience, yet this is accompanied by losing detail regarding deficits. We sought to determine whether a revised composite measure tailored to Michigan Medicaid beneficiaries would better identify those at risk for nursing home admission. We compared composite ADL measures created by exploratory factor analysis and additive modeling to Medicaid Enrollment, MI Choice Waiver program, and Nursing Facility claims data from 2013 to 2017. There were moderate to high levels of correlation between ADLs (.4-.82). Exploratory factor analysis extracted two factors, corresponding to domains of mobility or self-care tasks. Application of the self-care-based ADL limitations composite measure provided prediction power equivalent to an additive measure incorporating all ADL limitations for nursing home admission. This approach demonstrated improved interpretability with the need for just five measures.