Keynote Speakers

Anton Eremeev. Overcoming Local Optima in Evolutionary Heuristics: Theory and Practice

Shuai Li. Competition of Tribes and Cooperation of Members Algorithm: An Evolutionary Computation Approach for Model Free Optimization

Long Jin. Computational Social Science: Leveraging Data and Models to Unravel Opinion Dynamics and Drive Societal Change

Ying Liufu. Improved Neural Dynamics Models for Real-Time Nonlinear Optimization with Applications 

Mario Guarracino. Advances in Whole Graph Embedding: Techniques, Applications, and Performance Evaluation

Ratikanta Behera. Tensor computations with image applications

Overcoming Local Optima in Evolutionary Heuristics: Theory and Practice

One of the key questions in search for the global optimum in combinatorial optimization, non-convex mathematical programming and black-box optimization is how to overcome local optima and proceed towards a global optimum. Numerous approaches are developped to deal with this issue, and one of them is evolutionary algorithms (EAs), based on the principles of population search using selection, mutation and, sometimes, crossover operators. 

Recent results obtained in the theory of evolutionary algorithms provide rigorous analysis  of efficient and inefficient treatment of local optima by the EAs on certain illustrative families of problem instances in the search space of binary strings. These findings are supported by experimental results on some well-known benchmarks from combinatorial optimization. 

In this talk, the above mentioned results will be surveyed and complemented by the new results of the author and his colleagues, obtained for one simulation-optimization problem of buffers allocation in production lines.

Competition of Tribes and Cooperation of Members Algorithm: An Evolutionary Computation Approach for Model Free Optimization

Metaheuristic algorithms are a series of intelligent algorithms that are inspired by natural phenomenon and achieve the optimal searching via utilizing behaviors from nature. The rising complexity and dimensions of practical engineering problem in reality, a significant number of metaheuristic algorithms are promoted and applied in various of fields. Inspired by ancient tribes competition and members cooperative behavior, this paper proposes the Competition of Tribes and Cooperation of Members Algorithm (CTCM). Subsequent experiments are conducted on 23 benchmark test functions and exhaustively compared with other state-of-the-art algorithms, including particle swarm optimization (PSO), grey wolf optimizer (GWO), sparrow search algorithm (SSA), egret swarm optimization (ESOA), beetle antennae search (BAS) and whale optimization (WOA). The standard deviation and average, as well as statistic test are utilized to compare the performance of each algorithms, which reveal that CTCM outperforms in most kinds of problem. And from the result of Wilcoxon and Friedman rank test, the CTCM achieve the first place in all categories of problems, which indicate that CTCM possesses strong global optimization search capability and stability, and has faster convergence speed. The superiority of CTCM is then proofed on practical engineering optimization problems, in which CTCM achieve all the optimal solution for each engineering problem.

Computational Social Science: Leveraging Data and Models to Unravel Opinion Dynamics and Drive Societal Change

The digitization of information and advancements in computer technology have significantly impacted research methodologies in the field of social sciences, giving rise to a new and emerging discipline—Computational Social Science. This discipline utilizes computational models to explore complex social phenomena such as opinion dynamics and collective behavior. Professor Long Jin believes that "understanding interactions between individuals in social networks is key to comprehending how opinions evolve." In his presentation, he will focus on how he employs computational models, such as agent-based modeling, to study the spread of opinions, polarization, and consensus within social networks. Additionally, the presentation will cover how gaining deeper insights into the mechanisms of opinion formation can help address pressing societal challenges and, ultimately, exert broader influence on policy-making and social change.

Improved Neural Dynamics Models for Real-Time Nonlinear Optimization with Applications

Neural dynamics, a product at the intersection of artificial intelligence, dynamical systems, and control theory, characterizes system states as neurons and considers their dynamic evolution over time. When integrated with optimization problems transformed from physical systems, it forms a dynamical system imbued with physical knowledge, where constraints ensure system outputs adhere to physical laws. Over time, each neuron in neural dynamics progressively converges towards solutions for real-time optimization problems. In this keynote talk, Ying Liufu will elucidate the development of neural dynamics methods. Furthermore, the relevant work in real-time solving of nonlinear optimization problems using neural dynamics, delving into its application in various practical control systems such as redundant robots and autonomous vehicles.

Advances in Whole Graph Embedding: Techniques, Applications, and Performance Evaluation

Networks serve as effective models for capturing interactions and dependencies across various domains, from social networks to life sciences. These models can be extended to ensembles of networks, offering a versatile framework for modeling complex phenomena. Whole graph embedding is a powerful approach that projects entire graphs into a vector space, preserving their structural properties. In recent years, several embedding techniques have been introduced, including graph kernels, matrix factorization, and deep learning-based methods, all of which enable the creation of low-dimensional graph representations. These embeddings are useful for tasks such as feature extraction, graph clustering, and classification.

In this keynote speech, we will explore various techniques that jointly embed entire graphs for classification purposes. We will compare these methods and assess their performance on both synthetic and real-world undirected network datasets across different learning tasks.

Tensor computations with image applications

In the era of BIG data, artificial intelligence, and machine learning, we need to process multiway (tensor-shaped) data. These data are mainly in the three or higher-order dimensions, whose orders of magnitude can reach billions. Huge volumes of multidimensional data are a big challenge for processing and analyzing; the matrix representation of data analysis is not enough to represent all the information content of the multiway data in different fields. In this talk, we will discuss tensor factorization as a product of tensors. To address the factorizations, we define a closed multiplication operation between tensors with the concept of transpose, inverse, and identity of a tensor. We will conclude with a few color image applications in a tensor-structure domain.