Uplift Modeling aims to detect subgroups in a population with a specific response or reaction to an action taken on the targeted group. In these models, the Treatment set contains objects that have been exposed to some action, such as a marketing campaign or clinical treatment, while in the Control set, they have not. In this study, a novel ARTIFICIAL IMMUNE system-based model was designed using an AIRS classifier to solve uplift modeling problems with improved efficiency. In this approach, a predictive model was built for estimating the conditional probability of receiving the desired response from the subpopulation that has taken the action over the relevant probability of the sub-population that has not taken the action. The proposed model was tested on the Hillstorm-visit-w dataset. Experimental results showed a 138 percent improvement in the area under the uplift curve which is a measure to assess an uplift model's performance.

This article introduces two swarm intelligent ALGORITHMs, a group search optimizer (GSO) and an ARTIFICIAL fish swarm ALGORITHM (AFSA). A single intelligent ALGORITHM always has both merits in its specific formulation and deficiencies due to its inherent limitations. Therefore, we propose a mixture of these ALGORITHMs to create a new hybrid optimization ALGORITHM known as the group search-ARTIFICIAL fish swarm ALGORITHM (GS-AFSA). This ALGORITHM has been applied to three different discrete truss optimization problems. The optimization results are compared with those obtained using the standard GSO, the AFSA and the quick group search optimizer (QGSO). The proposed GS-AFSA eliminated the shortcomings of GSO regarding falling into the local optimum by taking advantage of AFSA’s stable convergence characteristics and achieving a better convergence rate and convergence accuracy than the GSO and the AFSA. Furthermore, the GS-AFSA has a superior convergence accuracy compared to the QGSO, all while solving a complicated structural optimization problem containing numerous design variables.