Abstract Various strategies for active learning have been proposed in the machine literature. In uncertainty sampling, which is among most popular approaches, learner sequentially queries label of those instances its current prediction maximally uncertain. The predictions as well measures used to quantify degree uncertainty, such entropy, are traditionally a probabilistic nature. Yet, alternative approaches capturing learning, alongside with corresponding measures, recent years. particular,...

<title>Abstract</title> In this study, we study the Variational Quantum Eigensolver (VQE) application for Ising model as a test bed model, in which pivotally delved into several optimization methods, both classical and quantum, analyzed quantum advantage that each of these methods offered, then proposed new combinatorial scheme, deemed QN-SPSA+PSR combines calculating approximately Fubini-study metric (QN-SPSA) exact evaluation gradient by Parameter-Shift Rule (PSR). The method integrates...

We propose a method for reliable prediction in multi-class classification, where reliability refers to the possibility of partial abstention cases uncertainty. More specifically, we allow predictions form preorder relations on set classes, thereby generalizing idea set-valued predictions. Our approach relies combining learning by pairwise comparison with recent proposal modeling uncertainty which distinction is made between reducible (a.k.a. epistemic) caused lack information and irreducible...

In this study, a large amount of fly ash (FA) and bottom (BA) from the Nghi Son coal-fired thermal power plant in Thanh Hoa, Vietnam is used producing unfired solid bricks. FA was utilized to substitute up 85% (by weight) cement amount, while BA as fine aggregate. Test results showed that all bricks produced study had unit weight 1.51 – 1.68 T/m3 were classified Grade M3.5 M15. As replacement level increased, weight, compressive strength, conductivity reduced, meanwhile water absorption...

In contrast to conventional (single-label) classification, the setting of multilabel classification (MLC) allows an instance belong several classes simultaneously. Thus, instead selecting a single class label, predictions take form subset all labels. this paper, we study extension MLC, in which learner is allowed partially abstain from prediction, that is, deliver on some but not necessarily We propose formalization MLC with abstention terms generalized loss minimization problem and present...

In contrast to conventional (single-label) classification, the setting of multilabel classification (MLC) allows an instance belong several classes simultaneously. Thus, instead selecting a single class label, predictions take form subset all labels. this paper, we study extension MLC, in which learner is allowed partially abstain from prediction, that is, deliver on some but not necessarily This option useful cases uncertainty, where does feel confident enough entire label set. Adopting...

When learning from instances whose output labels may be partial, the problem of knowing which these should made precise to improve accuracy predictions arises. This can seen as intersection two tasks: one partial and active learning, where goal is provide additional model accuracy. In this paper, we propose querying strategies for well-known K-nn classifier. We different criteria increasing complexity, using among other things amount ambiguity that introduce in decision rule. then show our...

In this study, we delved into several optimization methods, both classical and quantum, analyzed the quantum advantage that each of these methods offered, then proposed a new combinatorial scheme, deemed as QN-SPSA+PSR which combines calculating approximately Fubini-study metric (QN-SPSA) exact evaluation gradient by Parameter-Shift Rule (PSR). The method integrates QN-SPSA computational efficiency with precise computation PSR, improving stability convergence speed while maintaining low...