Contributions in Merlin

null

Biases impair the effectiveness of algorithms. For example, the age bias of the widely-used PageRank algorithm impairs its ability to effectively rank nodes in growing networks. PageRank’s temporal bias cannot be fully explained by existing analytic results that predict a linear relation between the expected PageRank score and the indegree of a given node. We show that in evolving networks, under a mean-field approximation, the expected PageRank score of a node can be expressed as the product of the node’s indegree and a previously-neglected age factor which can “amplify” the indegree’s age bias. We use two well-known empirical networks to show that our analytic results explain the observed PageRank’s age bias and, when there is an age bias amplification, they enable estimates of the node PageRank score that are more accurate than estimates based solely on local structural information. Accuracy gains are larger in degree-degree correlated networks, as revealed by a growing directed network model with tunable assortativity. Our approach can be used to analytically study other kinds of ranking bias.

null

Updates on ontologies affect the operations built on top of them. But not all changes are equal: some updates drastically change the result of operations; others lead to minor variations, if any. Hence, estimating the impact of a change ex-ante is highly important, as it might make ontology engineers aware of the consequences of their action during editing. However, in order to estimate the impact of changes, we need to understand how to measure them. To address this gap for embeddings, we propose a new measure called \textit{Embedding Resemblance Indicator} (ERI), which takes into account both the stochasticity of learning embeddings as well as the shortcomings of established comparison methods. We base ERI on (i) a similarity score, (ii) a robustness factor $\hat{\mu}$ (based on the embedding method, similarity measure, and dataset), and (iii) the number of added or deleted entities to the embedding computed with the Jaccard index. To evaluate ERI, we investigate its usage in the context of two biomedical ontologies and three embedding methods---GraRep, LINE, and DeepWalk---as well as the two standard benchmark datasets---FB15k-237 and Wordnet-18-RR---with TransE and RESCAL embeddings. To study different aspects of ERI, we introduce synthetic changes in the knowledge graphs, generating two test-cases with five versions each and compare their impact with the expected behaviour. Our studies suggests that ERI behaves as expected and captures the similarity of embeddings based on the severity of changes. ERI is crucial for enabling further studies into impact of changes on embeddings.

This paper establishes existence of subgame perfect equilibrium in pure strategies for a general class of sequential multi-lateral bargaining games, without assuming a stationary setting. The only required hypothesis is that utility functions are continuous on the space of economic outcomes. In particular, no assumption on the space of feasible payoffs is needed. The result covers arbitrary and even time-varying bargaining protocols (acceptance rules), externalities, and other-regarding preferences. As a side result, we clarify the meaning of assumptions on “continuity at infinity.”

Exponential Random Graph Models (ERGMs) have gained increasing popularity over the years. Rooted into statistical physics, the ERGMs framework has been successfully employed for reconstructing networks, detecting statistically significant patterns in graphs, counting networked configurations with given properties. From a technical point of view, the ERGMs workflow is defined by two subsequent optimization steps: the first one concerns the maximization of Shannon entropy and leads to identify the functional form of the ensemble probability distribution that is maximally non-committal with respect to the missing information; the second one concerns the maximization of the likelihood function induced by this probability distribution and leads to its numerical determination. This second step translates into the resolution of a system of O(N) non-linear, coupled equations (with N being the total number of nodes of the network under analysis), a problem that is affected by three main issues, i.e. accuracy, speed and scalability. The present paper aims at addressing these problems by comparing the performance of three algorithms (i.e. Newton’s method, a quasi-Newton method and a recently-proposed fixed-point recipe) in solving several ERGMs, defined by binary and weighted constraints in both a directed and an undirected fashion. While Newton’s method performs best for relatively little networks, the fixed-point recipe is to be preferred when large configurations are considered, as it ensures convergence to the solution within seconds for networks with hundreds of thousands of nodes (e.g. the Internet, Bitcoin). We attach to the paper a Python code implementing the three aforementioned algorithms on all the ERGMs considered in the present work.

null

null

In recent years, there has been a growing interest in social and emotional skills (SES) both in the scientific literature and in social practice. The paper presents an overview of the ways of understanding what SES are and the catalogs thereof. There are some attempts in the literature to organize these catalogs within the Big Five traits that for a long time was claimed to be the most sound model of basic orthogonal dimensions of personality. However, further research on personality structure revealed that two metatraits can be found above the Big Five traits. These two metatraits form the basis of the Two Factor Model of personality, which was later developed into the Circumplex of Personality Metatraits. It turned out that in certain aspects models based on metatraits have a greater theoretical potential than those based on the Big Five traits. The paper presents a proposal for describing SES from the perspective of the Circumplex of Personality Metatraits rather than the Big Five. In this framework, we distinguish the concept of personality competences that underlie and organize many specific SES and identify the core personality competencies on the basis of the Circumplex of Personality Metatraits model.

Theoretical accounts distinguish between motivational ('wanting') and hedonic ('liking') dimensions of rewards. Previous animal and human research linked wanting and liking to anatomically and neurochemically distinct brain mechanisms, but it remains unknown how the different brain regions and neurotransmitter systems interact in processing distinct reward dimensions. Here, we assessed how pharmacological manipulations of opioid and dopamine receptor activation modulate the neural processing of wanting and liking in humans in a randomized, placebo-controlled, double-blind clinical trial. Reducing opioid receptor activation with naltrexone selectively reduced wanting of rewards, which on a neural level was reflected by stronger coupling between dorsolateral prefrontal cortex and the striatum under naltrexone compared with placebo. In contrast, reducing dopaminergic neurotransmission with amisulpride revealed no robust effects on behavior or neural activity. Our findings thus provide insights into how opioid receptors mediate neural connectivity related to specifically motivational, not hedonic, aspects of rewards.