Research - Ran Cheng

Dr. Ran Cheng   

Ph.D., Research Fellow

CERCIA Group, School of Computer Science,
University of Birmingham,
United Kingdom, B15 2TT

Email: ranchengcn (at) gmail.com
Curriculum Vitae, Google Scholar, Research Gate

[Home] [Research] [Publications] [Professional Activities] [Honors and Awards]


With the development of modern engineering designs and scientific research, optimization problems become increasingly complex. Most of these complex optimization problems can no longer be solved using traditional mathematical tools. Alternatively, there is an increasing interest in adopting computational intelligence (CI) techniques (a set of nature-inspired computational methodologies and approaches) to address these problems.

I am particularly interested in CI based optimization of complex real-world problems, e.g., large optimization problems. Related research topics include swarm intelligence, model based evolutionary algorithms, evolutionary multi-objective optimization, evolutionary multi-modal optimization, etc.

Without loss of generality, a minimization (or maximization) optimization problem can involve one or multiple objectives to be optimized, which can be formulated in a uniform manner as:

\[
\begin{aligned}
\min\limits_{\mathbf{x}} \quad & \mathbf{f}(\mathbf{x}) = (f_1(\mathbf{x}), f_2(\mathbf{x}), ..., f_M(\mathbf{x}) ), \\
\text{s.t.} \quad & \mathbf{x} \in X, \quad \mathbf{f} \in Y,\\
& g_i(\mathbf{x}) \le 0, i = 1,...,P \\
& h_j(\mathbf{x}) = 0, j = 1,...,Q \\
\end{aligned}
\label{eq:problem}
\]
where \(X \subseteq \mathbb{R}^D\) is the decision space and \(\mathbf{x} = (x_1, x_2, ..., x_D) \in X\) is the decision vector; \(Y \subseteq \mathbb{R}^M\) is the objective space and \(\mathbf{f} \in Y\) is the objective vector. The decision vector \(\mathbf{x}\) is composed of \(D\) decision variables while the objective vector \(\mathbf{f}\) is composed of \(M\) objective functions that map \(\mathbf{x}\) from \(X\) to \(Y\). \(g_i(\mathbf{x}) \) and \(h_j(\mathbf{x}) \) are known as the inequality and equality constraints respectively. If the number of objective functions is only one, i.e., \(M = 1\), the problem is often known as a single-objective optimization problem (SOP); while if there is more than one objective function in conflict with each other, i.e., \(M > 1\), the problem is often known as a multi-objective optimization problems (MOP). Specially, if an MOP has more than three objectives, it is known as an many-objective optimization problem (MaOP) due to the particular challenges.

Large optimization problems generally refer to those involving a large number of decision variables and/or many objectives. To be specific, large optimization problems can be categorized into the following four types:

It is worth noting that, L-SOPs are also known as large-scale optimization problems for short in the literature of single-objective optimization, which is, however, different from the definition as presented here.

Swarm Intelligence


Swarm intelligence (SI) is the collective behavior of decentralized, self-organized systems, natural or artificial. Representative SI paradigms include particle swarm optimization, ant colony optimization, bee algorithms, etc.

Model Based Evolutionary Algorithms


Model based evolutionary algorithms (MBEAs) generally refer to machine learning related techniques applied to evolutionary computation, such as theories, algorithms, systems and applications. Related topics include estimation of distribution algorithms, Bayesian optimization algorithms, surrogate-assisted evolutionary computation, data-driven optimization, inverse modelling for multi-objective optimization, etc.

Evolutionary Multi-objective Optimization


Evolutionary multi-objective optimization (EMO) generally refers to the methodologies of applying population based metaheuristics (e.g., EAs) to solving MOPs (or MaOPs). Related topics include many-objective optimization, preference articulation/integration and decision-making, constraint handling, etc.

Evolutionary Multi-modal Optimization


Multimodal optimization (MMO), which refers to single-objective optimization involving multiple optimal (or near-optimal) solutions, has attracted increasing interest recently. MMO is widely seen in real-world scenarios, where the decision-makings can be made on the basis of multiple optimal solutions of a given optimization problem. For example, in truss-structure optimization, where the optimization objective is the quality criterion (e.g. weight or reliability) of the truss structure and the decision variables can be the density or length of the truss members, it is likely that different values of the decision variables can lead to the same (or very close) fitness of the objective function. In such a scenario, the decision maker (DM) has to make decisions according to personal preferences.

References


  1. Ran Cheng. Nature Inspired Optimization of Large Problems, University of Surrey, UK, May 2016.
  2. Ran Cheng and Yaochu Jin. A Competitive Swarm Optimizer for Large Scale Optimization. IEEE Transactions on Cybernetics, 45(2): 191-204, 2015. (Top 1% ESI Highly Cited Article) [PDF] [Matlab Code] [C Code]
  3. Ran Cheng and Yaochu Jin. A Social Learning Particle Swarm Optimization Algorithm for Scalable Optimization. Information Sciences, 291: 43-60, 2015. (Top 1% ESI Highly Cited Article) [PDF] [Matlab Code] [C Code]
  4. Ran Cheng , Yaochu Jin, Kaname Narukawa and Bernhard Sendhoff. A Multiobjective Evolutionary Algorithm using Gaussian Process based Inverse Modeling. IEEE Transactions on Evolutionary Computation, 19(6): 838-856, 2015. [PDF] [Matlab Code]
  5. Ran Cheng, Yaochu Jin, Markus Olhofer and Bernhard Sendhoff. A Reference Vector Guided Evolutionary Algorithm for Many-objective Optimization. IEEE Transactions on Evolutionary Computation, 20(5): 773-791, 2016. (Top 5 Popular Article of IEEE Transactions on Evolutionary Computation in October 2016) [PDF] [Matlab Code] [Java Code]
  6. Ran Cheng, Tobias Rodemann, Michael Fischer, Markus Olhofer, and Yaochu Jin. Evolutionary Many-objective Optimization of Hybrid Electric Vehicle Control: From General Optimization to Preference Articulation. IEEE Transactions on Emerging Topics in Computational Intelligence, 2017. (in press) [PDF]
  7. Ran Cheng, Yaochu Jin, Markus Olhofer and Bernhard Sendhoff. Test Problems for Large-scale Multiobjective and Many-objective Optimization. IEEE Transactions on Cybernetics, 47(12): 4108-4121, 2017. [PDF] [Matlab Code]
  8. Xingyi Zhang, Ye Tian, Ran Cheng* (Corresponding Author), and Yaochu Jin. A Decision Variable Clustering-Based Evolutionary Algorithm for Large-scale Many-objective Optimization. IEEE Transactions on Evolutionary Computation, 2016. (in press) [PDF] [Matlab Code]