Swarm Intelligence

Swarm Intelligence Overview

Swarm Intelligence:

1) One Million Heads, One Beautiful Mind

2) Agents interacting locally with each other and the environment

3) Agents follow simple rules

4) Emergence of Itelligent, Collective, Self-organised, Global behaviour

5) Decentralized and artificial or natural

6) Very adaptive

7) Randomness enables the continuous exploration of the alternatives and it ensures that the better solution will be found.

8) Behavior relies on stochastic choices made by the agents which are a balance between a simple perception-reaction model and a random model.

9) Application of bio-inspired concepts

10) Large mass of the agents is a must.

Swarm Intelligence Capabilities:

1) Scheduling / Load Balancing: The emphasis is on the relative position of the job rather than its direct predecessor or its direct successor in the schedule and summation evaluation rule / global pheromone evaluation rule is followed.

Fig: Ant Colony Load Balancing - AntZ

2) Clustering: A cluster is a collection of agents which are similar and are dissimilar to the agents in other clusters.

Fig: Formation of cluster of corpses to clean up the ants' nests

3) Optimization: An optimization problem is the problem of finding the Best Solution / Minimal Cost Solution from all the feasible solutions.

Fig: optical network optimization - A practical application of particle swarm intelligence

4) Routing: This is based on the principle that backward ants utilize the useful information gathered by the forward ants on their trip from source to destination.

Fig: The AntHocNet routing algorithm for MANETs (mobile ad hoc networks)

Advantages:

1) Flexible: The colony respond to internal disturbances and external challenges.

2) Robust: Tasks are completed even if some agents fail.

3) Scalable: From a few agents to millions

4) Decentralized: There is no central control in the colony.

5) Self-organized: The solutions are emergent rather than pre-defined.

6) Adaptation: The swarm system can not only adjust to predetermined stimuli but also to new stimuli.

7) Speed: Changes in the network can be propagated very fast.

8) Modularity: Agents act independently of other network layers.

9) Parallelism: Agents' operations are inherently parallel.

Disadvantages:

1) Behaviour: Difficult to predict the behaviour from the individual rules.

2) Knowledge: The functions of colony could not be understood with the knowledge of functioning of a agent.

3) Sensitivity: Even a small change in the simple rules results in different group level behaviour.

4) Action: Agent behavior looks like noise as action of choice is stochastic.

First level Swarm Intelligence:

It uses a Positive Feedback Pheromone to mark low-distance routes with a entry signal for their colleagues.

Second level Swarm Intelligence:

It uses a Negative Pheromone to mark unrewarding paths with a no-entry signal.

Exploration vs Exploitation: Important Factors in the design of the Optimization Techniques

1) In swarm intelligence systems, Exploration Phase introduces the freedom and Exploitation Phase introduces the usage of collected search experience to locate the optimal solutions without wasting too much time.

2) Two new techniques have been introduced for Particle Swarm Optimization:

Resets increase exploitation
Delayed Updates increase exploration

Swarmic Freedom vs Random Freedom:

1) The freedom of the swarm (i.e. Swarmic Freedom) is maintained by the swarm intelligence algorithms, whereas the freedom of the agents (i.e. Random Freedom) is maintained by the randomised algorithms.

2) Swarm Freedom naturally enables the system to maintain recognisable ﬁdelity to the original input whereas when more freedom is granted to the randomised algorithm, the algorithm soon begins to deviate excessively.

The Bottom-Up Strategy for Engineering Emergent Behavior:

This strategy in swarm system builds complex social simulations by designing agents with individual self-organization, psychologically reasonable behaviors to generate consequential social event / emergent behaviour.

Weak vs. Strong Computational Creativity In Swarm Intelligence Systems:

1) In Strong Computational Creativity, artificial intelligence is strong whereas in Weak Computational Creativity, artificial intelligence is weak.

2) In Strong Computational Creativity, the machines are expected to be creative and to have genuine understanding and other cognitive states as well as capability of conscious thinking and therefore the stress for the emergent creativity is on the signiﬁcant impact of both freedom and constraint whereas Weak Computational Creativity does not go beyond exploring the simulation of human creativity.

3) Swarm intelligence via its infamous struggle identifies a suitable balance between exploration and exploitation phases to provide a valuable mean artiﬁcial creativity approach.

General Swarm Principles:

1) Proximity principle: The basic units of a swarm should be capable of giving the respond back to to environmental variance triggered by interactions among agents. However, some fundamental behaviors are shared such as living-resource searching and nest-building.

2) Quality principle: A swarm should be able to respond to quality factors such as determining the safety of a location.

3) Principle of diverse response: Resources should not be concentrated in a narrow region. The distribution should be designed so that each agent will be maximally protected facing environmental fluctuations.

4) Principle of stability: The population should not change its mode of behavior every time the environment changes.

5) Principle of adaptability: The swarm is sensitive to the changes in the environment that result in different swarm behaviour.

Most Common Principles:

1) Individuals are attracted to each other.

2) When they come closer in space, they start to focus in the same direction.

3) They avoid collision by moving away from each other. They keep certain distance from each other.

4) Individuals interact with local, near neighbours and trust only a few of them. This rule is known as self-organization.

Business Swarm:

1) Swarm intelligence can be used by any business or social cause. It could give you cost efficiency advantage.

2) Swarm Intelligence emerges when individuals unite to a cause.

3) Individuals unite to a cause when the goal is clearly defined, immediately seen as achievable, touches the individuals who are ready to jump in it without asking anyone's permission and electrify individuals to shoot.

4) Your oganization must be optimized for speed, trust and scalability.

5) Build a scafffolding that supports the swarms.

6) Your organization must be decentralized with small distributed responsibilities.

7) Different approaches must be carried out in parallel.

8) Make the environment full of fun.

Swarm Intelligence Aspects:

Distribution:

Agents choose their actions and then carry them out.

Stigmergy:

1) Agents indirectly interact via environmental modification, the phenomenon known as stigmergy.

2) Stigmergy is basically the context awareness.

3) Stigmergy decouples agents' interactions.

Cooperation:

1) Agents cooperate to emerge a non-deterministic, complex collective behaviour.

2) Agents cooperate in order to solve complex tasks.

Self Organization:
The bases of self-organization are:

1) Positive feedback (amplification): It promotes better solutions by allocating to them more agents.

2) Negative feedback (for counter-balance and stabilization): It may avoid that all individuals converge to the same behavior or to the same state.

3) Amplification of fluctuations (randomness, errors, random walks)

4) Multiple social interactions

5) There is a continuous tension between positive feedback and negative feedback and this is what actually happens in most known self-organization phenomena, e.g., cellular automata, markets, complex networks, etc.

Emergence:

1) Complicated intelligent behaviour emerges from simple agents following simple rules.

2) Weak Emergence: You can trace the agent behaviour from emergent properties.

3) Strong Emergence: Agent behaviour is not directly traceable from emergent behaviour.

Imitates Nature:

Artificial swarm is designed by imitating the natural swarm behaviour.

General swarm behaviour:

1) Foraging: To search for the food

2) To construct the nest

3) To move in the environment

Swarm Machinery / Swarm Inteligence Mechanisms:

A Swarm Machinery is all about Agents, Interactions and the Environment.

Environmental mechanism:

Agents indirectly interact with each other via environmental modification which serves as external memory. The phenomenon is known as Stigmergy which means stimulation by work.

Interaction mechanism:

There are no direct communications but indirect interactions between the agents via environmental modification known as stigmergy / stimulation by work.

Activities of agents

1) Action is stochastic choice based, a balance between a simple perception-reaction model and a random model

2) Agents react on the basis of simple perception-reaction model according to which individuals perceive the local properties of the environment and also affect the properties of the environment to some extent.

3) Agents move in environment

Swarm Technologies:

Particle Swarm Optimization:

Inspiration: Particle Swarm Optimization is inspired by the social foraging behavior of some animals such as flocking behavior of birds and the schooling behavior of fishes.

Strategy: The goal of the algorithm is to have all the particles locate the optima in a multi-dimensional space, initially assigned with random position and random velocity, gradually advancing towards the local optima through the use of exploration and exploitation of good, known positions in space.

PSO Rules:

1) Separation: Do not run into flockmates.

2) Alignment: Each align their own heading to the average of the neighbours.

3) Cohesion: Move towards the average position of neighbours.

4) Desire factor per bird for roosting areas: It is the need of roosting or swarming which gets stronger if a defined roosting area is approachable.

Ant System:

Inspiration: It is inspired by the pheromone communication of the blind ants regarding a good path between colony and the food source in an environment, the phenomenon known as stigmergy. The probability of the ant following a certain route is not only a function of pheromone intensity but also a function of distance to that city, the function known as visiblity.

Strategy: The objective of the strategy is to exploit historic i.e. pheromone based and heuristic information to construct candidate solutions each in a probabilistic step-wise manner and fold the information learned from constructing solutions into the history. The probability of selecting a component is determined by the heuristic contribution of the component to the overall cost of the solution and the quality of solution and history is updated proportional to the quality of the best known solution.

Bees Algorithm:

Inspiration: It is inspired by the foraging behaviour of the honey bees. The hive sends out the Scout bees which when locate nectar (a sugary fluid secreted within flowers), return to the hive and communicate the other bees the fitness, the quality, distance and direction of the food source via waggle dance.

Strategy: The objective of the algorithm is to locate and explore good sites within a problem search space. Many scout bees are sent out, each iteration is always in search of additional good sites which are continually exploited in the a local search application.

Fireworks Algorithm:

Inspiration: It is inspired by observing the firework explosion.

Strategy: In the FA, two explosion (search) processes are employed and mechanisms for keeping the diversity of sparks are also well designed. The explosion process of a firework can be viewed as a search in the local space around a specific point where the firework is set off through the sparks generated in the explosion.

Fig: State transition in lifecycle model of bacteria in BFO

Bacterial Foraging Optimization Algorithm:

Fig: Chemotactic behavior of E. coli - run and tumble

Inspiration: It is inspired by the the foraging behavior of E.coli bacteria that will perceive chemical gradients in the environment (such as nutrients) and move towards or away from the specific signals.

Strategy: The objective of the algorithm is to allow cells to stochastically and collectively swarm toward optima through a series of three processes:

Chemotaxis: Here the cost of the cells is inversely proportional to the proximity to other cells and cells move along manipulated cost surface area one at a time.
Reproduction: Only those cells contribute to this phase who remain healthiest in their overall life-time.
Elimination-dispersal: Some cells are discarded and new cells are inserted.

River Formation Dynamics Algorithm:

Inspiration: RFD is inspired by how water forms rivers by eroding the ground and depositing sediments. As water transforms the environment, altitudes of places are dynamically modified, and decreasing gradients are constructed. The gradients are followed by subsequent drops to create new gradients, reinforcing the best ones. By doing so, good solutions are given in the form of decreasing altitudes.

Metaphor: A set of drops placed at the starting point is subjected to gravitational force that attracts them to the center of the earth. As a result, these drops are distributed throughout their environment, seeking the lowest point or the sea. As a result, the riverbeds are formed, often containing many meanders.

Strategy: RFD utilizes this idea into the graph theory problems (for example, the problems of finding a minimum distance tree and finding a minimum spanning tree in a variable-cost graph), creating a set of agent-droplets moving on the edges between nodes according to the decreasing gradient of the nodes and exploring the environment for the best solution following the mechanisms of erosion and soil sedimentation that relate to the altitude assigned to each node.

Intelligent Water Drops:

Inspiration: Intelligent water drops algorithm (IWD) inspired by natural rivers and how they find almost optimal paths to their destination. These near optimal or optimal paths follow from actions and reactions occurring among the water drops and the water drops with their riverbeds.

Metaphor: This is achieved by three important characteristics of water drops. First, they have a velocity that allows them to gather soil from the river bed, thus the higher the speed of drops, the larger the amount of soil it carries. In this way the water drop cleanse the path for the forthcoming drops. Second, the velocity of water drops increases more on paths with minimal soil than on one with the high soil. Third, when a single water drop has to select a path, it selects the one with the lowest amount of soil.

Strategy: Several artificial water drops cooperate to change their environment in such a way that the optimal path is revealed as the one with the lowest soil on its links. The solutions are incrementally constructed by the IWD algorithm.

Gravitational search algorithm:

Inspiration: Gravitational search algorithm (GSA) is a newly developed stochastic search algorithm based on the Newtonian gravity- "Every particle in the universe attracts every other particle with a force that is directly proportional to the product of their masses and inversely proportional to the square of the distance between them" and the mass interactions.

Strategy: In this approach, the search agents are a collection of masses which interact with each other based on the Newtonian gravity and the laws of motion in which all of the objects attract each other by the gravity force, while this force causes a global movement of all objects towards the objects with heavier masses. The heavy masses correspond to good solutions of the problem.

Ant based Clustering Algorithm:

1) ACO is finding the shortest way by the ants and ACO clustering is the finding the shortest way between the data items of a given data-set to be clustered. For example: ACO based documents clustering is finding the most alike(the shortest way between the documents) documents.

2) Seen purely as a clustering algorithm, ant-based clustering performs well in comparison to the other popular clustering methods of k-means, agglomerative hierarchical clustering and one dimensional self-organising maps.

SoS-ACO (Sense of Smell - Ant Colony Optimization): A Bio-inspired algorithm for searching Relationships in Social Networks.

1) It accelerates the search for relationships among elements present in social networks.

2) It involves locating the chain of reference that leads from one person to another by accelerating the search for routes between two nodes that belong to a graph that represents a social network.

3) SoS-ACO is based on the way the ants move when they search for food.

4) The application of this algorithm to real social networks obtains an optimal response in a very short time (tens of milliseconds).

Shuffled Frog-Leaping

Metaphor: The SFL algorithm involves a population of possible solutions deﬁned by a set of frogs (i.e. solutions) that is partitioned into subsets referred to as memeplexes. The different memeplexes are considered as different cultures of frogs, each performing a local search. Within each memeplex, the individual frog holds ideas, that can be inﬂuenced by the ideas of other frogs, and evolve through a process of memetic evolution. After a number of memetic evolution steps, ideas are passed among memeplexes in a shuffling process. The local search and the shuffling processes continue until convergence criteria are satisﬁed.

Cuckoo Search:

CS is an optimization algorithm inspired by the obligate blood parasitism of some cuckoo species by laying their eggs in the nests of other host birds.

Methaphor: Cuckoos have an aggressive reproduction strategy that involves the female laying her fertilized eggs in the nests of other species so that surrogate parents accidently raise her brood. Sometimes the cuckoo's egg in the host nest is discovered, the surrogate parents either throw it out or abandon the nest and builds their own brood elsewhere but otherwise once the first cuckoo chick is hatched, it first evicts the host eggs by blindly propelling the eggs out of the nest so that its share of food is increased.

Swarm Languages:

Protoswarm:

1) It is a programming Language for programming Multi-robot System Using the Amorphous Medium Abstraction.

2) Inspiration: It is inspired by the continuous space-time model of Proto and extends this type of model to program swarm of robots.

3) Amorphous Medium Abstraction is achieved using two mechanisms: a language, called Protoswarm, which provides continuous space and time semantics, and a runtime library which approximates the semantics on the given hardware.

Swarm:

1) It is a true distributed programming language.

2) The Fundamental Concept behind Swarm: We should move the computation, not the data.

3) The Swarm Prototype: It is a stack-based language, similar to a primitive version of the Java bytecode interpreter and is now implemented as a Scala library.

StarLogo:

1) A programming language and programming environment of decentralized multi-agent system.

2) StarLogo is a programmable modeling environment for exploring the working of decentralized systems - systems which are organized without an organizer, coordinated without a coordinator.

3) With StarLogo, you can model (and gain insights into) many real-life phenomena, such as bird flocks, traffic jams, ant colonies, and market economies.

Growing Point Language:

1) It is a programming language to program amorphous computing medium to generate highly complex and prespecified patterns.

2) It is a programming language that facilitates the self-organisation of complex pre-specified patterns, such as the interconnection structure of an arbitrary electrical circuit.

3) Inspiration: It is inspired by a botanical metaphor: a growing point is a locus of activity that can propagate through the amorphous computing medium by transferring its activity from one computing element to a neighbouring one, according to the growing point's tropism (pheromone).

Swarm Frameworks:

ASSL (Autonomic System Specification Language):

1) ASSL is a framework for formally specifying and generating autonomic systems.

2) AS are formal executable models with an interaction protocol and autonomic elements.

3) In an endeavor to specify ANTS (Autonomous Nano-Technology Swarm) with ASSL, we have emphasized modeling ANTS' self-management policies such as self-configuring, self-healing, self-scheduling, and emergent self-adapting. In addition, we have developed specification models for the ANTS safety requirements.

Swarm Organ: A theoretical framework for organising the swarm of Gene Regulatory Network-controlled agents which display Adaptive Tissue like organization.

DDDAS (Dynamic Data Driven Application System) for Artificial Swarm Control:

1) It is the ability to dynamically incorporate additional data into an executing application, and in reverse, ability of an application to dynamically steer the measurement process.

2) Two application designs have been proposed to control the swarm application via DDDAS: Swarm Application Architecture integrated with DDDAS and DDDAS Swarm Control Framework.
3) Swarm Application Architecture:

Here a swarm application has been integrated with DDDAS to control several agents by broadcasting one or few swarm parameters and then report application performance to the central controller as a single, aggregated statistic.

4) DDDAS Swarm Control Framework: DDDAS framework with feedback control loop for swarm control allows the controller to appropriately adjust agents' parameters for the application.

This framework incorporates the swarm application architecture.
It facilitates improved analytic, predictive, and decision-making capabilities of swarm application by Synergestic Feedback Loop which is achieved by executing the same swarm application in multiple parallel simulations, each utilize the same real data with different agent-level control parameters.

A Distributed Framework for Supporting 3D Swarming Applications:

1) It supports swarming within in-flight sensor networks (swarm of quad copters in this study), which require coordinated movement in 3D space.

2) High impact 3D swarming applications include autonomous mapping, surveying, servicing, environmental monitoring and disaster site management.

3) A general hardware setting for controlling swarms of in-flight sensor networks has been proposed by combining IoT (Internet of Things) technology with swarm robotics.

A General Modeling Framework for Swarm:

Inspiration:
1) In the majority of the research on swarm intelligent systems, the moving agents in the swarm are modeled as simple reactive agents. Current swarm models comprise a little intelligence to fully exploit the potential of swarms.
2) The analysis of a swarm intelligence system typically focuses on the dynamics of the swarm as a whole, rather than on the dynamics of the individual agents.

Concept:

The most prevalent control problem in swarm literature: Aggregation and coordinated motion of the swarm-bot are studied in this reasearch.
This framework uses two major swarm intelligence methods in optimization and control, namely Particle Swarm Optimization (PSO) and Artiﬁcial Potential Fields for Swarm Aggregation.

A Unified Framework for Swarm Based Image Analysis:

1) It is only a proposal and currently there is no way of implementing swarm algorithm for image ananlysis.

2) The main goal is to achieve a global perception of one image as the emergent sum of local perceptions of the whole colony.

3) Unlike ACO and PSO, this focuses on constructing different rules and interactions for obtaining a specific emergent behavior, which is then used in an image analysis context.

Figure: No direct association among a peer index and the keys stored by the peer with each key is stored in a peer whose centroid is very close to that key value

Self-Chord: It is a Bio-inspired P2P Framework for Self-organized, Distributed Grid Information System.

1) Self-Chord is a P2P system that inherits Chord-like structured systems for the construction and maintenance of an overlay of peers, but derives the autonomy behavior, self-organization and capacity to adapt to a changing environment features from ant-inspired algorithms for key ordering and discovery.

2) Peer indexes and resource keys are uncorrelated, which opens the possibility to give a semantic meaning to keys and perform class (or range) queries.

3) Mobile agents go through the Chord ring and order resource keys.

4) Benefits: 1) Dynamic behavior (management of perturbations, such as the connection of new peers), 2) Load balancing, 3) Possibility to serve range queries, as the keys can be assigned semantic values.

So-Grid: It is a Bio-Inspired P2P Framework for Self-organized, Distributed Grid Information System.

1) A set of bio-inspired algorithms tailored to the decentralized construction of a Grid information system that features adaptive and self-organization characteristics.

2) So-Grid provides two main functionalities which is being exploited continuously and concurrently:

Logical Reorganization of Resources, inspired by the behavior of ants and termites, is done by moving and collecting the items within their environment and then spatially sorting resource descriptors over the Grid according to their classiﬁcation.
Resource Discovery, inspired by the ants mechanism in which they search for food sources by following the pheromone traces left by other ants.

Antare:
1) It is an Ant-Inspired P2P Information System for a Self-Structured Grid.
2) It is designed to effectively disseminate and reorganize resources to speed up resource discovery operations in a dynamically changing environment.

Myra: A cross-platform Ant Colony Optimization framework written in Java

AntCar: A program for solving the car sequencing problem.

The Swarm Framework:

1) The ultimate Platform as a Service used for the distributed computation in the cloud.

It is going much further than systems like Google App Engine in relieving the programmer from the difficulties of cloud computing.

Swarm Tools for Optimization and Feature Extraction:

The Swarm Application Framework (SAF):
A tool that allows the engineer to design swarm applications from the top down, so the design problem becomes more manageable.

A prototype of the framework:

The goals of SAF are:

To make the development of swarm applications intuitive.
To allow the engineer to design swarm applications from the abstract (swarm) level instead of the individual (agent) level.
To modify the swarm behavior with top-level rules instead of modifying multiple low-level rules.
To make rule abstraction a simple process, thus promoting the use of rule hierarchies.
To provide a collection of modules that developers can use to quickly build new applications.
To enable the engineers to focus on the behavior and properties of the swarm, rather than on the low-level details of an agent behavior.
To create the swarm applications via the Rule abstraction mechanism in SAF is much easier than creating them from scratch.

Route Optimization of Unmanned Aerial Vehicles (UAV):

UAV is an aircraft without the onboard presence of pilots.
It includes software and hardware agents that communicate or displace in an optimal manner.
The UAVs are engaged in a simulated area coverage scenario with a defined set of waypoints. The objective is to find the shortest route that connects all the waypoints in order to optimize the time and the cost of the UAV’s flight.
Ant system algorithm is used for UAVs route optimization.
It is being highly used in commercial aplications such as telecommunications, ground traffic control, search and rescue operations, and crop monitoring among others.
UAVs assist with frost protection, irrigation and crop management in agriculture.
Together with Mobile Ground Station systems, UAVs offer persistent surveillance, enhanced situational awareness, and actionable intelligence to the law enforcement and the security personnel.

Ant System-based Adaptive Edge Detector:

Two algorithms- Ant System Algorithm for Edge Detection and Ant System-based Broken-edge Linking Algorithm inspired by the foraging behavior of natural ant colonies are used.
Ant System Algorithm for Edge Detection: This method requires that a set of images is extracted from the original grayscale image using a nonlinear image enhancement technique called Multiscale Adaptive Gain, and then the modified AS algorithm is applied to detect the edges on each of the extracted images.
Ant System-based Broken-edge Linking Algorithm: It is complementary to edge detection and is used to connect the broken edges in order to form the closed contours (outline) that separate the regions of interest.
This model is created with a bottom-up approach, using the rules of local interactions between the ants and the environment (digital image). This model is decentralized, self-organized, autonomous and adaptive to the changes in the environment.

Distributed Task Allocation in Large, Autonomous, Multirobot Swarm System:

Distributed Bees Algorithm (DBA) is used which is inspired by the foraging behaviour of colonies of bees in nature.
When the bees find the food source, the scout bees return to the hive and perform a famous waggle dance in order to recruit other bees. The information about the richness and location of the source is passed using direct communication. In the same way, here the robots are designed to use broadcast communication to inform other robots in the range about the estimated location and the quality of the found target.
The objective of the proposed algorithm is to assign the robots in a swarm to the found targets in such a way that the final distribution is proportional to the targets’ qualities.
Distributed Bees Algorithm provides the robot swarm with scalability in terms of the number of robots and number of targets and with adaptability to a non-uniform distribution of the targets’ qualities.
The bottom-up design topology inherent to bio-inspired multirobot systems provides them with one or more of the following features, such as being autonomous, scalable, robust and adaptive to changes in their environment.

Figure: Comparison of the segmentation results for a ROI mammogram, 256 × 256 pixels. The ASCA (Ant System-based Cluster Algorithm) extracted ﬁve clusters;

Cluster Analysis for Image Segmentation:

Image segmentation is an important preprocessing step in applications of computer vision. The objective is to partition the image into homogeneous regions that share certain visual characteristics.
ASCA (Ant System-based Clustering Algorithm) is used which consists of three consecutive parts, namely:
a) Pheromone accumulation, b) Local pheromone summing, and c) Data labeling
a) Pheromone accumulation is used to create a pheromone map of the data set to be clustered and b) Local pheromone summing in which smooth pheromone surface is obtained by locally summing all the pheromone trails and c) Data labeling in which all the nodes are grouped in their respective cluster and the all the clusters are extracted from the data set.
It is used to extract the pixel clusters with a similar intensity level of grey.
This tool is used in computer vision applications such as mammography for the cancer risk analysis and breast cancer where the less representative pixels are precisely the most interesting because they represent a variation with respect to healthy tissue.
The important feature of the proposed ASCA algorithm is automatic extraction of clusters.
ASCA algorithm outperformed 1D-SOM, k-Means, FCM and PFCM algorithms in detection of small, atypical regions of the image.

Use cases:

Possible applications of Swarm Intelligence may be limited only by imagination.

Behavioral Animation:

The particle swarm technology concepts are being applied in computer graphics area and can be found in Batman Returns (1992), The Lion King (1994) and From Dusk Till Dawn (1996).
The most impressive usage are probably the immense battle sequences in the trilogy Lord of the Rings where about 250,000 individual fighters.
For making this possible a new software was written named MASSIVE which controls this mass of agent technology-equipped computer actors and their states.

Distributed Perceptive Networks - These are the examples of emergence of intelligence and artificial swarm intelligence in artificial complex systems.

A system comprising hundreds or thousands of motes linked by radio transceivers and sensors can spontaneously emerge as a perceptive network and a mote is a micromachine which is the unit of SmartDust, each unit installed with TinyOS.
Multi-hop networking approach is followed and as a result, parallelism is achieved so that if a particular mote stops functioning, there is enough redundancy and parallelism in the network that other motes reconfigure the connectivity to bypass that mote.
The technology of placing the brain, the sensors, and the actuators of an artificial intelligent structure in differnt locations known as Pervasive Computing.

Micro-satellite Swarm:

Bluetronix is looking to develop control packages and communications suites founded upon swarm intelligence algorithms enabling collaboration of micro-satellite swarms, tasking of individuals, and fuzzy system identification for adaptive sensor fusion dictating rule based commands.
Individual satellites, each equipped with their own rule-based controller, will perform assigned sub-tasks based on their own directives.
This design could be easily integrated on a wide variety of mobile platforms including satellites, Earth Science sensor networks, ground stations, and small aircraft, all connected in an ad-hoc fashion.

Bluetronix Swarm Mobile Ad Hoc Network:

The Swarm Autonomous Routing Algorithm (SARA) is performed by simple communication node devices for node to node communications in a network, especially a Mobile Ad hoc NETwork (MANET). These networks are decentralized with the ability to scale to 100s to thousands of connections. They also self-learn and organize as they operate and adapt as new and old nodes enter and exit the network under dynamic conditions.

Robust, Flexible, Easy-to-use, Swarm Energy Management:

REGEN Energy’s wireless automated demand management and demand response controllers can be easily installed onto any electrical heating, cooling or discretionary electrical load. Once installed, the REGEN controllers work together like a swarm of bees, intelligently communicate and manage the duty cycles of the loads being controlled. Utilizing REGEN’s patented swarm-based intelligence, the controllers dramatically reduce peak electrical demand by up to 25% in commercial and light industrial properties and allow for effective scheduling of overnight and weekend loads.

Fig: Swarm Energy Management employs swarm logic to allow equipments in buildings to communicate and coordinate to minimize the number and size of loads unnecessarily running concurrently, thereby reducing peak demand.

SLAM with PSO: - Simultaneous Localization and Mapping with Particle Swarm Optimization

Estimate the pose (position and orientation) of a robot and map the environment at the same time.
Learning a map and locating the robot simultaneously.
Localization: Inferring the location given a map.
Mapping: Inferring a map given the locations of robot.
FastSLAM is a framework for simultaneous localization using a Rao-Blackwellized particle filter. In FastSLAM, particle filter is used for the mobile robot pose (position and orientation) estimation, and an Extended Kalman Filter (EKF) is used for the feature location’s estimation. However, FastSLAM degenerates over time.
A Neuro-Fuzzy Multi Swarm FastSLAM Framework both Extended kalman filter for landmark feature estimation, and a particle filter based on particle swarm optimization are presented to overcome the impoverishment of FastSLAM.

Figure: The Swarm Planning Framework - Here the landscape operates as a swarm, so when the external pressure moves them away from the equilibrium, then it simply shifts to another stable state through its self-organization capability and this regime shift is visualized as a ﬁtness landscape.

Swarm Planning: The development of a planning methodology to deal with climate adaptation.

Swarm planning is a theoretical and practical approach to deal with the uncertain future.
Swarm planning theory is used in two pilot designs: Post-carbon world and Pre-adaptive landscape and is compared with regular planning process. The results are presented in the form of two new landscapes: the Zero-Fossil Region, where the design provides a spatial framework for a complete renewable energy supply,
and the Net Carbon Capture Landscape, in which adaptation and mitigation strategies are designed to become carbon positive.
The comparison illuminates the potential advantages of swarm planning to tackle the climate change threats.
It increases the ﬂexibility of spatial systems in two ways: Assist the change in spatial land use over time; and catalyse the emergence of autonomous and more resilient developments.

Swarm Robotics:

Swarm Robotics is the study of how large number of relatively simple physically embodied agents can be designed such that a desired collective behaviour emerges from the local interactions among agents and between the agents and the environment. It has some special characteristics, which are found in swarms of insects, that is, decentralised control, lack of synchronisation, simple and quasi identical members.

Fig: The DAG given above depicts a different model in which the alarm will ring when activated by high temperature and/or coolant water pipe leakage in the reactor.

Bayesian Network Structure Learning:

Bayesian Network / Belief Network is a probabilistic directed acyclic graphical model that represents a set of random variables and their conditional dependencies.
A Bayesian network could represent the probabilistic relationship between diseases and symptoms. Given symptoms, the network can be used to compute the probability of the presence of various diseases.
Bayesian networks that model the sequence of variables (e.g. speech signals or protein sequences) are called dynamic Bayesian networks.
Two novel approaches (ChainACO and K2ACO) based on chain structure model and K2 greedy search are being used for for Bayesian network structure learning.
It consists of two phases: Construct chains (i.e. contruct the order of nodes according to dependencies), Apply K2ACO to the best ordering found and returns the best structure.
Example: Given a node ordering X1,X2,. . . ,Xn, we deﬁne the chain structure by adding edges between successive nodes. Thus Xi is the sole parent of Xi+1. Ei is the edge from Xi to Xi+1
ACO-based Bayesian network learning algorithm outperforms greedy search and simulated annealing algorithms.

From Fireflies to Fault Tolerant Swarm:

It is a decentralized system that detects non-operational robots in the swarm by engineering a flash light system on the robots.
This flash light system is similar to some firefies species which can synchronize their flashing.
This robotic approach creates the ability for operational robots to flash in unison; failed robots can thus be detected as those that will not flash in synchrony with the rest of the robot team.

SINS: Sound INterfacing through the Swarm

Fundamental concept of acoustic:

Audio INterface is generally the hardware that connects your microphones and other audio gear to your computer. A typical audio interface converts analog signal into the digital audio information that your computer can process.
Acoustic sensing is generally the sense of hearing.
In SINS swarm system, every device will have a wireless connection, hence leading to trillions of connected devices and sensors known as the the sensory swarm.
The SINS performance and lifetime objectives are to achieve high-resolution acoustic activity detection and beyond 1-meter accurate acoustic localization, while achieving above 1-year lifetime on 1 button cell battery.

Swarmanoid: Towards Humanoid Robotic Swarms

The Swarmanoid is an autonomous group of approximately 60 robots who work together like a SWAT (acronym for "Special Weapons And Tactics") team to accomplish the most mundane tasks.
This machine swarm is made up of flying eye-bots, gripping hand-bots, and wheeled foot-bots.
The main scientific objective of this research project is the design, implementation and control of a novel distributed robotic system.

Flight of the Robobee: The Rise of Swarm Robotics

RoboBee is a tiny robot capable of tethered flight, developed by a research robotics team at Harvard University.
A swarm of robotic bees
RoboBees will be deployed on search and rescue missions, or used for military surveillance.
RoboBees could even pollinate crops and flowers, thus replicate the behavior of their biological cousins.

ANTS / PAM: Autonomous Nanotechnological Swarm - Mission Architecture / Prospecting Asteroid Mission - ANTS application

The primary objective of PAM is the exploration of the asteroid belt in search of resources and materials with astrobiologically relevant origins and signatures.
PAM plans to drive a carrier spaceship and have it self-assemble and launch 1000 small exploration spacecrafts (picocrafts) that are to travel through and analyze asteroid belt.
Each spacecraft includes a team leader (ruler), one or more messengers, and a number of workers.
The messengers are needed to connect the team members when they cannot connect directly, due to a long distance or a barrier.
Once launched, spacecraft opportunistically self-organise into several sub-swarms and simultaneously analyse different asteroids over the several years belt traversal.
Each sub-swarm can repeatedly search for, detect and navigate towards interesting asteroid targets; mesaure and create 3D models of analysed asteroids; send adequate asteroid models to an Earth center.

Swarm Tile: Human interface tiles that are designed to understand their environment and react to various conditions.

The Cellular Intelligence system is a network constructed of modular and human interface tiles that are designed to understand their environment and react to various conditions.
Each cell or module is composed of smart sensing technology and one or multiple LED actuators.
Strategy: Each cell evaluates its surrounding environment through sensing technology or through network connections. If a stimulus is present, the information is processed using an arduino microcontroller, which is an open-source electronics prototyping platform, and an appropriate response is sent back to the cell itself or to another cell in the system and actuation is carried out through LED lights.

Fig: NANOMA based breast cancer therapy

Fig: A drug delivery vector

NANOMA: Nano-Actuators and Nano-sensors for Medical Applications:

NANOMA aims at developing drug delivery microrobotic systems composed of nanoActuators and nanoSensors for the propulsion and navigation of ferromagnetic microcapsules in the cardiovascular system through the induction on magnetic gradients.
New approach for breast cancer therapy based on Nanoma concept:
1. MRI based detection and tracking
2. MRI based in-vivo propulsion and navigation
3. Targeted drug delivery using functionalized nanovectors

HiveOS Network Operating System:

HiveOS enables Aerohive devices to organize into groups, or hives, which allows functionality like fast roaming, user-based access control and fully stateful firewall policies, as well as additional security and RF networking features - all without the need for a centralized or dedicated controller.
HiveOS has two primary feature sets - Wi-fi features and Routing features
All Aerohive devices support the feature-rich HiveOS Cooperative Control architecture.
Aerohive provides a pure cloud-enabled management solution for your wired and wireless network designed by Apple Education Experts.

Nanorobots / Nanobots / DNA nanotechnology in Medical Applications:

It constitutes any smart structure capable of actuation, sensing, signaling, information processing, intelligence, manipulation and nano scale (10-9 m) swarm behaviour.
Nanorobots could propose solutions at the most of the nanomedicine problems.
NanoRobotics – An Example: Ultra-local Drug Delivery
The technology is known as DNA origami, or alternately DNA nanotechnology.

Fig: Spectral Biclustering of Microarray Data: Coclustering Genes and Conditions

Biclustering of Microarray Gene Expression Data:

Gene expression is the process by which information from a gene is used in the synthesis of a functional gene product often termed as protein.
In Gene regulatory network, genes have been regarded as nodes in a network, with inputs being proteins such as transcription factors, and outputs being the level of gene expression.
Microarray is a 2D array on a solid substrate. Microarray gene expression data is a 2D array of gene expression data under some condition. Microarray gene expression data plays a vital role in biological processes, gene regulation and disease mechanisms.
In the Gene Expression context, clustering is defined as the grouping of genes based on the similarity of their condition feature profile whereas the biclustering finds subsets of genes that show similar patterns under a specific subset of experimental conditions or in short it finds regulatory patterns. For example: for microarray data, strong up-regulation of certain genes under a cancer condition of a particular type. Such a simultaneous classiﬁcation of samples and features is called biclustering (or co-clustering).
The algorithmic concepts of the Particle Swarm Optimization (PSO), Shuffled Frog Leaping (SFL) and Cuckoo Search (CS) algorithms have been analyzed for the four benchmark gene expression dataset and the experiment results show that CS outperforms PSO and SFL for 3 datasets and SFL give better performance in one dataset.

Swarm Intelligence Approach for Accurate Gene Selection in DNA Microarrays:

It is the challenge of extracting the specific genes responsible for the given illness.
The goal is to minimise classification errors whilst using the smallest possible set of genes to explain the results provided by the given DNA microarray.
This model consists of a particle swarm optimization (PSO) algorithm, in which a feature selection mechanism facilitates identification of small samples of informative genes among thousands of genes.

Computer-aided Detection of Breast Cancer on Mammograms: A Swarm Intelligence Optimized Wavelet Neural Network Approach

The proposed abnormality detection algorithm is based on extracting Laws Texture Energy Measures from the mammograms and classifying the suspicious regions by applying a pattern classifier.
The method has been applied to real clinical database of 216 mammograms collected from mammogram screening centers.

Potential of Swarm Intelligence in Big Data Analytics:

The main focus is on data.
The other three properties of big data analytics, which include the high dimensionality of data, the dynamical change of data, and the multi-objective problems.
Based on the combination of swarm intelligence and data mining techniques, we can have better understanding of the big data analytics problems, and then we can design more effective algorithms to solve real-world big data analytics problems.

Machine Learning Tools and Particle Swarm Optimization for Content-Based Search in Big Multimedia Databases:

Here PSO's role is in data clustering, image retrieval and function minimization in solving this problem.

Swarm Intelligence for Trafﬁc Light Scheduling:

In this context, our main objective is to ﬁnd optimized cycle programs (OCP) for all the trafﬁc lights located in a given urban area.
Speciﬁcally, cycle programs are refereed to the time span a set of trafﬁc lights (in a junction) keep their color states.
At the same time, these programs have to coordinate trafﬁc lights in adjacent intersections with the aim of improving the global ﬂow of vehicles circulating according to trafﬁc regulations.

AntNet: Ant-based Swarm Intelligence Algorithm for Routing in Communication Networks

It is an improvement of ant-based algorithms achieved via dynamic programming.
In the AntNet algorithm, routing is determined by means of very complex interactions of forward and backward network exploration agents (ants).
The idea behind this sub division of agents is to allow the backward ants to utilize the useful information gathered by the forward ants on their trip from source to destination.
Based on this principle, no node routing updates are performed by the forward ants. Their only purpose in life is to report network delay conditions to the backward ants, in the form of trip times between each network node. The backward ants inherit this raw data and use it to update the routing table of the nodes.

Swarm Intelligence

References: