Fast incremental learning of logistic model tree using least angle regression

Lee, Sudong

doi:10.1016/j.eswa.2017.12.014

Cited by 34 publications

(10 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The measure of performance between the algorithms is provided by the AUC, which indicates how much a model is capable of distinguishing between classes: a model whose predictions are 100% wrong has an AUC of 0.0; one whose predictions are 100% correct has an AUC of 1.0. We compare the performance of the iSVM with two well-known incremental algorithms, the incremental Logistic Regression (iLR) [37] and an incremental artificial neural network, the incremental Radial Basis Function network (iRBF) [38]. The testing network used for the validation is composed by a requester interacting with nodes, as providers, that implement each a different behaviour, from benevolent to all of the seven possible attacks.…”

Section: B Simulation Results For ML Algorithmsmentioning

confidence: 99%

Trust-Related Attacks and Their Detection: A Trust Management Model for the Social IoT

Marche

Nitti

2021

IEEE Trans. Netw. Serv. Manage.

View full text Add to dashboard Cite

The integration of social networking concepts into the Internet of Things (IoT) has led to the so called Social Internet of Things paradigm, according to which the objects are capable of establishing social relationships in an autonomous way with respect to their owners. Within this scenario, "things" interact opportunistically with their peers to seek needed services. However, attacks and malfunctions in the IoT can outweigh any of its benefits if not handled adequately. In this paper, we focus on the possible types of trust attacks that can affect the IoT and propose a trust management model able to overcome all the analyzed attacks. Simulations show how the proposed model can effectively isolate almost any malicious nodes in the network at the expense of an increase in the number of transactions needed for the model to converge.

show abstract

Section: B Simulation Results For ML Algorithmsmentioning

confidence: 99%

Trust-Related Attacks and Their Detection: A Trust Management Model for the Social IoT

Marche

Nitti

2021

IEEE Trans. Netw. Serv. Manage.

View full text Add to dashboard Cite

show abstract

“…It is known through the implantation process of Z‐yarns that applying tension to Z‐yarns can make the friction angle with the fibers in X and Y directions close to 90° when entering the preform, thus reducing the wear of fibers. [ 21 ] A tension control platform was designed, as shown in Figure 8, to achieve tension regulation by installing weights on one side of the fiber, but due to the frictional effect of the weights and pulleys, the weight of the weights cannot be equated to the tension, so a digital acquisition module was used to convert the weight of the weights into tension, as shown in Figure 9, the weight of the weights was F 1 , and the force F 1 ′ was generated by using a tension sensor to clamp one end of the Z‐yarn and drag it horizontally. Tension under different weights were recorded by the digital acquisition module, and the results are shown in Table 1.…”

Section: Z‐yarn Wear Experimentsmentioning

confidence: 99%

Study on friction and wear of preform Z‐directional fiber

et al. 2022

View full text Add to dashboard Cite

In order to reduce the wear and tear of Z-yarn implantation into the preform, the frictional behavior of Z-yarn during implantation into the preform was studied. According to the actual implantation conditions of Z-yarns, we designed a guiding array clamping device and a tensile force sensor to study the changes of tensile strength of Z-yarns at different implantation positions and implantation times, results show that the tensile strength of Z-yarns decreases significantly after 25 implantations at the position of 4 N friction force, and it's necessary to replace Z-yarns in time to avoid their fracture in the preform. The use of unadulterated solvent dimethyl silicone oil applied to the Z-yarns, can significantly reduce their wear, the tensile strength of Z-yarns after 60 implantations is only 5.6% lower than that of Z-yarns after 10 implantations. Design the tension control system, and when the tension is 0.46 N, the tensile strength of Z-yarns during implantation is 34.2% higher than without tension. Meanwhile, the relationship between the normal force applied to Zyarn and the implantation length and fiber bundle width are established by combining Hertzian theory and experimental data, and the friction coefficient algorithm during Z-yarn implantation is derived by combining Howell's equation, which solves the problem that the fiber friction coefficient cannot be measured due to the complicated working condition of Z-yarn implantation.The frictional wear mechanism of yarn during Z-yarn implantation into the preform is revealed, and a method to reduce yarn wear and increase its continuous implantation number is proposed.

show abstract

“…Lee and Jun investigated the computational overhead for an updated version of a logistic model tree (LMT) [13]. LMT complements logistic regression with tree induction to harness the advantages of both algorithms.…”

Section: B Incremental and Non-incremental Learningmentioning

confidence: 99%