An Efficient Approach to Detecting Lung Nodules Using Swin Transformer

Lung cancer has the highest rate of cancer-caused deaths, and early-stage diagnosis could increase the survival rate. Lung nodules are common indicators of lung cancer, making their detection crucial. Various lung nodule detection models exist, but many lack efficiency. Hence, we propose a more efficient approach by leveraging 2D CT slices, reducing computational load and complexity in training and inference. We employ the tiny version of Swin Transformer to benefit from Vision Transformers (ViT) while maintaining low computational cost. A Feature Pyramid Network is added to enhance detection, particularly for small nodules. Additionally, Transfer Learning is used to accelerate training. Our experimental results show that the proposed model outperforms state-of-the-art methods, achieving higher mAP and mAR for small nodules by 1.3% and 1.6%, respectively. Overall, our model achieves the highest mAP of 94.7% and mAR of 94.9%.

Solving Minimum Vertex Cover Problem Using Learning Automata

Minimum vertex cover problem is an NP-Hard problem with the aim of finding minimum number of vertices to cover graph. In this paper, a learning automaton based algorithm is proposed to find minimum vertex cover in graph. In the proposed algorithm, each vertex of graph is equipped with a learning automaton that has two actions in the candidate or non-candidate of the corresponding vertex cover set. Due to characteristics of learning automata, this algorithm significantly reduces the number of covering vertices of graph. The proposed algorithm based on learning automata iteratively minimize the candidate vertex cover through the update its action probability. As the proposed algorithm proceeds, a candidate solution nears to optimal solution of the minimum vertex cover problem. In order to evaluate the proposed algorithm, several experiments conducted on DIMACS dataset which compared to conventional methods. Experimental results show the major superiority of the proposed algorithm over the other methods.

Finding a Maximum Clique using Ant Colony Optimization and Particle Swarm Optimization in Social Networks

Interaction between users in online social networks plays a key role in social network analysis. One on important types of social group is full connected relation between some users, which known as clique structure. Therefore finding a maximum clique is essential for some analysis. In this paper, we proposed a new method using ant colony optimization algorithm and particle swarm optimization algorithm. In the proposed method, in order to attain better results, it is improved process of pheromone update by particle swarm optimization. Simulation results on popular standard social network benchmarks in comparison standard ant colony optimization algorithm are shown a relative enhancement of proposed algorithm.

A Multi-Swarm Cellular PSO based on Clonal Selection Algorithm in Dynamic Environments

Many real-world problems are dynamic optimization problems. In this case, the optima in the environment change dynamically. Therefore, traditional optimization algorithms disable to track and find optima. In this paper, a new multi-swarm cellular particle swarm optimization based on clonal selection algorithm (CPSOC) is proposed for dynamic environments. In the proposed algorithm, the search space is partitioned into cells by a cellular automaton. Clustered particles in each cell, which make a sub-swarm, are evolved by the particle swarm optimization and clonal selection algorithm. Experimental results on Moving Peaks Benchmark demonstrate the superiority of the CPSOC its popular methods.

Tracking Extrema in Dynamic Environment using Multi-Swarm Cellular PSO with Local Search

Many real-world phenomena can be modelled as dynamic optimization problems. In such cases, the environment problem changes dynamically and therefore, conventional methods are not capable of dealing with such problems. In this paper, a novel multi-swarm cellular particle swarm optimization algorithm is proposed by clustering and local search. In the proposed algorithm, the search space is partitioned into cells, while the particles identify changes in the search space and form clusters to create sub-swarms. Then a local search is applied to improve the solutions in the each cell. Simulation results for static standard benchmarks and dynamic environments show superiority of the proposed method over other alternative approaches.