Special Session on Nature-Inspired Algorithms and Machine Learning
Machine learning techniques have been the forerunner for several application domains that are driven by decision-making mechanisms. Despite the outstanding results in a broad range of applications, deep neural networks are usually treated as black-boxes since it is not straightforward to design automatic approaches that can either learn hyperparameters/parameters or proper architectures. Most of the so-called end-to-end techniques do not consider parameter learning, and they are restricted to a few applications only. Therefore, to learn and proper design pattern classifiers and feature extractors can be modeled as an optimization problem.
On the other hand, nature-inspired optimization techniques have been employed in several areas of knowledge, but not yet fully considered for designing deep networks or even traditional pattern classifiers. The high-dimensional feature maps constructed by deep learning models can be further refined using feature selection or other feature transformation step, and then used as an input to classifiers that are known to be sensitive to sparse feature vectors but with good generalization capabilities in so many other situations.
Besides, research on natural language processing together with nature-inspired techniques and deep models are yet to be explored in their full content, thus being an interesting niche of research with a vast number of applications. It is very often to find networks with hundreds of thousands of parameters to be learned, which is usually costly and derivative-sensitive.
The above reasons motivated us to propose this special section on nature-inspired techniques applied to machine learning in general. The event aims at fostering discussion and debate among participants around the theoretical analysis, implementation details, and case studies related to the application of nature-inspired optimization algorithms to problems related to the main topic of the Special Session.
Scope: The main topics of interest include (not limited to):
- Fine-tuning hyperparameters/parameters in neural networks and pattern classifiers.
- Learning architectures in deep networks.
- Feature selection, learning, and weighting.
- Learning cost functions in adversarial networks.
- Hybrid nature-inspired metaheuristics applied to the design of classifiers.
- Ensemble learning.
- Multi and many-objective optimization applied to machine learning.
- Evolutionary concept drift.
- Nature-inspired regularization techniques for deep neural networks.
Please follow the submission guideline from the IEEE CEC 2021 Submission Website. Special session papers are treated the same as regular conference papers. Please specify that your paper is for the Special Session on Nature Inspired Algorithms and Machine Learning. All papers accepted and presented at CEC 2021 will be included in the conference proceedings published by IEEE Explore, which are typically indexed by EI.
- 31 January 2021: Paper Submission Deadline
- 2 March 2021: Paper Acceptance Notification
- 7 April 2021: Final Paper Submission & Early Registration Deadline
- 28 June – 1 July 2021: Conference Dates
- João Paulo Papa
- School of Sciences, São Paulo State University, Bauru, Brazil.
- E-mail: firstname.lastname@example.org
- Xin-She Yang
- School of Science and Technology, Middlesex University, London, UK.
- E-mail: email@example.com