This article is a part of a series of articles. Links to previous parts can be found in the appendix section of the article.
Krill Herd Optimization (KH) Algorithm
Krill are tiny crustaceans found in almost all of the world’s oceans. In the ocean food chain, Krill is a critical trophic level connection. They are near the bottom of the food chain in the ocean. They feed on phytoplankton and zooplankton and multiply quickly, converting these abundant resources into a form suitable for many larger ocean animals. For many of these larger animals, krills make up the most significant part of their diets
Krills play a critical role in the ocean ecosystem, and therefore, over the last few decades, extensive studies have been conducted to understand the ecology and distribution of the krill population. Although there are still uncertainties around the drivers determining the distribution of the krill herds, many conceptual models have been formulated to explain the observed formation of the krill herds. The core results obtained by such conceptual models reveal that the krill swarms form the basic unit of organization for their population.
Hence, it becomes critical to understand the formation of the krill swarms and the causes and factors leveraged by krill as adaptive advantages of aggregation formation. When predators, such as seals, penguins, or seabirds, attack krill, they eliminate individual krill. This results in reducing the krill density of the herd. The herd then repopulates itself by following a specific behavior. The repopulation of the herd after a hunting event depends on various parameters. The herding process of krill individuals essentially has two main goals (objectives):
- Increasing krill density in the herd
- Reaching nearest good food source
After studying the herding behavior of the krill in seas, Gandomi and Alavi proposed a new swarm intelligence-based global optimization algorithm in 2012 called krill herd (KH) optimization. This algorithm captures the multi-objective herding process in a new metaheuristic algorithm for solving global optimization problems.
Taking cues from the herding behavior, density-dependent attraction of krill (increasing density of the herd) and finding food (finding areas of high food concentration) are used as objectives that finally lead the krill to herd around the global minima. In this process of herding, an individual krill moves toward the best solution when it searches for the highest density and food. This means that the closer the distance to the high density and food, the lesser the objective function. Note that the time-dependent position of an individual krill in the two-dimensional surface is determined by three main actions :
- Movement induced by other individuals in the herd
- Foraging action
- Random diffusion
Applications of Krill Herd Algorithm
A consistent theme is that swarm intelligence algorithms are a great fit for power economic dispatch optimization tasks. You can find some examples below. As highlighted in this series, this application can be extrapolated for end-to-end real-time supply chain flow optimization. Not something that most companies are geared to implement currently, but is undoubtedly the future of supply chain optimization and planning.
- Application of bio-inspired krill herd algorithm to combined heat and power economic dispatch
- Economic load dispatch using krill herd algorithm
- Krill Herd algorithm solution for the economic emission load dispatch in power system operations
An exciting application is designing supply chain networks. A neural network paired optimization algorithm can help design supply chain networks using the most granular data, which is currently very challenging in network optimization. Large organizations need to aggregate inputs like products to eliminate the granularity. While our rationale is that it does not impact the results much, that is true only in specific scenarios. We will cover this in detail in the coming week’s episode of “Think About It.” If you read an example paper (below), you can easily correlate how it can be extrapolated to supply chain network design.
Application of Fine-Tuned Krill Herd Algorithm in Design of Water Distribution Networks
Another area where these algorithms find good application is problems that leverage classification. This paper provides a good overview of how this algorithm can be leveraged in classification. While an overview, you can imagine the plethora of ways this algorithm can be leveraged in classification problems as you read the paper.
An Enhanced Krill Herd Optimization Technique Used for Classification Problem
Some additional research papers highlighting the classification capabilities have been listed below.
- Krill Herd Optimization algorithm for cancer feature selection and random forest technique for classification
- Hybrid Monkey Algorithm with Krill Herd Algorithm optimization for feature selection
- A Krill Herd Algorithm For Efficient Text Documents Clustering
You have probably already noticed the last one in the list above. This algorithm can help enhance AI-based document management and contract management in the supply chain context. Paired with AI-based approaches like NLP, this algorithm can better classify documents than the k-means approach.
We will continue our journey of swarm intelligence algorithms in the subsequent article. The article will be published on 12/11.
Appendix
This article is part of a series of articles. Previous parts of this article series can be found here:
Designed Analytics Blog 