The increasing complexity of real-world optimization problems demands fast, robust, and efficient meta-heuristic algorithms. The popularity of these intelligent techniques is gaining popularity day by day among researchers from various disciplines of science and engineering. The sine cosine algorithm is a simple population-based stochastic approach for handling different optimization problems. In this work, we have discussed the basic sine cosine algorithm for continuous optimization problems, the multi-objective sine cosine algorithm for handling multi-objective optimization problems, and the discrete (or, binary) versions of sine cosine algorithm for discrete optimization problems. Sine cosine algorithm (SCA) has reportedly shown competitive results when compared to other meta-heuristic algorithms. The easy implementation and less number of parameters make the SCA algorithm, a recommended choice for performing various optimization tasks. In this present chapter, we have studied different modifications and strategies for the advancement of the sine cosine algorithm. The incorporation of concepts like opposition-based learning, quantum simulation, and hybridization with other meta-heuristic algorithms have increased the efficiency and robustness of the SCA algorithm, and meanwhile, these techniques have also increased the application spectrum of the sine cosine algorithm. The integration of machine learning techniques with the sine cosine algorithm is also becoming a hot topic for researchers in various fields. For instance, Zamli et al. incorporated the concept of Q-learning along with the levy flights in the sine cosine algorithm (QL-SCA) for improving the performance of the algorithm [1]. Wang et al. combined the mechanism of extreme learning machines (ELM) and sine cosine algorithm to develop a novel air quality prediction model [2]. Fu et al. utilized the sine cosine algorithm with variational mode decomposition (VMD) model, and singular value decomposition (SVD)-based phase space reconstruction (PSR) method to optimize least squares support vector machine (LSSVM) [3]. Similarly, Sahlol et al. utilized the sine cosine algorithm (SCA) for the training of multi-layer perceptron (MLP) or feedforward neural networks (FNNs) to improve the prediction of liver enzymes on fish farmed on nano-selenite [4]. The SCA algorithm is used in the training phase of NN to update the weights and the biases of the network. Hamdan et al. used the sine cosine algorithm (SCA) to train an artificial neural network (ANN) for forecasting the electricity demand [5]. And, Song et al. utilized sine cosine algorithm (SCA) to optimize the classification performance of the back-propagation neural network (BP-NN) for the image classification task [6]. The weights of the BP-NN were optimized using the SCA in the training phase to improve the classification accuracy.

Despite having all the characteristics of a good optimizer, the sine cosine algorithm still needs more attention from the researchers and focused application-oriented approaches in the development. The incorporation of various advance strategies and modifications can enhance the performance and optimization capabilities of the sine cosine algorithm.