Figure 2 - uploaded by Michael Christoph Thrun
Content may be subject to copyright.
Source publication
Algorithms implementing populations of agents which interact with one another and sense their environment may exhibit emergent behavior such as self-organization and swarm intelligence. Here a swarm system, called Databionic swarm (DBS), is introduced which is able to adapt itself to structures of high-dimensional data characterized by distance and...
Contexts in source publication
Context 1
... the literature, the existence of emergence is controversial 2 ; it is possible that the concept is only required because the causal explanations for certain phenomena have not yet been found [Janich/Duncker, 2011, p. 23]. Figure 2.1 presents an example of emergence in swarms. The non-deterministic movement of fish is temporarily and structurally unpredictable and consists of many interactions among many agents. ...
Context 2
... For the gross-domestic-product of 160 countries, the first cluster consists of mostly African and Asian countries and a second cluster consists of mostly European and American countries [Thrun, 2019]. Different types of distance and density-based high-dimensional structures are shown in Fig. A2, A.3 and A.4 where DBS is able to preserve them. Using DBS, new knowledge was extracted out of high-dimensional structures with regards to genetic data and multivariate time series (Fig. A5, ...
Citations
... HABC-MBOA was found to be useful in preventing sensor node overloading when used as CH. Thrun et al. 31 presented the databionic swarm system. As a result, while Swarm-based algorithms are task-centred and data independent approaches for WSN optimization issues they struggle with premature convergence problems leading to suboptimal performance in Heterogeneous networks due to their static approach of deployment irrespective the network dynamics and heterogeneous energy-based node implementation 32 . ...
... Improving network performance requires minimizing fitness F in Eq. (31). Comprises a wide range of fitness metrics derived by the provided Eqs. ...
... (26), (27), (28), (29), (30). Parameters utilized in the integration of the fitness function are given varying degrees of importance according to the weight coefficients in Eq. (31). It is up to the user to fine-tune these settings for their specific sensor network deployment. ...
Energy efficiency plays a major role in sustaining lifespan and stability of the network, being one of most critical factors in wireless sensor networks (WSNs). To overcome the problem of energy depletion in WSN, this paper proposes a new Energy Efficient Clustering Scheme named African Vulture Optimization Algorithm based EECS (AVOACS) using AVOA. The proposed AVOACS method improves clustering by including four critical terms: communication mode decider, distance of sink and nodes, residual energy and intra-cluster distance. Through mimicking the natural scavenging behavior of African vultures, AVOACS continuously balances energy consumption on nodes resulting in an increase in network stability and lifetime. For CH selection, we use AVOACS, which considers the following parameters: communication mode decider, the distance between sink and node, residual energy, and intra-cluster distance. In comparison to the OE2-LB protocol, simulation findings demonstrate that AVOACS enhances stability, network lifetime, and throughput by 21.5%, 31.4%, and 16.9%, respectively. The results show that AVOACS is an effective clustering algorithm for energy-efficient operation in heterogeneous WSN environments as it contributes to a large increase of network lifetime and significant enhancement of performance.
... In sum, DBS is a nonlinear projection that displays the structure of the highdimensional data into a low-dimensional space, preserving the cluster structure of the data. This model exploits the concepts of self-organization and emergence, game theory and swarm intelligence (Thrun, 2018;Thrun and Ultsch, 2021). Pswarm does not require any input parameters other than the dataset of interest and is able to adapt itself to structures of high-dimensional data such as natural clusters characterized by distance and/or density-based structures in the data space. ...
This Element highlights the employment within archaeology of classification methods developed in the field of chemometrics, artificial intelligence, and Bayesian statistics. These run in both high- and low-dimensional environments and often have better results than traditional methods. Instead of a theoretical approach, it provides examples of how to apply these methods to real data using lithic and ceramic archaeological materials as case studies. A detailed explanation of how to process data in R (The R Project for Statistical Computing), as well as the respective code, are also provided in this Element.
... SI techniques mimic the collective behavior of natural organisms, such as the flocking of birds, the schooling of fish, or the foraging of bees, to search for optimal solutions to complex problems [11]. SI techniques have been widely applied across diverse domains, including Engineering, Finance, Computer Science, and Social Sciences [12]. ...
... Let xk = wk be the beamforming vector to be found using the PSO algorithm, which is computed based on the cost function of (12). We use the projection method [22] to solve the optimization problem with constraint (12) by defining the feasible region F as follows: ...
This study delves into the energy optimization problem in Internet of Things (IoT) networks. We consider the downlink from multiple antenna Gateway (GW) and single antenna IoT devices. For this challenging nonconvex problem, we initially introduced the well-known zero-forcing beamforming (ZFBF) to eliminate inter-user interference, thereby transforming the energy efficiency maximization problem into a concave-convex fractional problem. Then, instead of applying a combination of ZFBF with power allocation, we propose the Particle Swarm Optimization (PSO) algorithm to allocate power to find the optimized beamforming matrix. Through extensive numerical analysis, we demonstrate the effectiveness of our proposed scheme in terms of energy efficiency and power achieved at the GW. The results underscore the significant benefits of our approach over conventional methods, paving the way for practical and efficient energy optimization in IoT networks.
... Swarm intelligence is governed by several key principles, including decentralization, self-organization, robustness, adaptability, and scalability [14]. Decentralization refers to the absence of a central controller, with individuals making decisions based on local information and interactions [15]. Self-organization arises from the interactions between agents, leading to the emergence of collective behaviours without external intervention [16]. ...
... Additionally, self-organization promotes scalability and flexibility, as swarm dynamics can scale seamlessly from small to large populations of agents without requiring explicit coordination or communication overhead [19]. By leveraging the self-organizing capabilities of swarm intelligence, complex tasks can be decomposed into simpler subproblems distributed among individual agents, thereby enhancing problem-solving efficiency and resource utilization [15]. Table 1 2019 This systematic review paper likely examines the application of swarm intelligence techniques for clustering tasks in data mining, providing insights into algorithmic approaches and their performance in clustering applications. ...
... Thrun & Ultsch [15] 2021 This paper likely explores the application of swarm intelligence techniques for selforganized clustering tasks, discussing algorithms, approaches, and their effectiveness in forming clusters without centralized control. ...
Swarm intelligence, inspired by the collective behaviour of natural swarms and social insects, represents a powerful paradigm for solving complex optimization and decision-making problems. In this review paper, we provide an overview of swarm intelligence, covering its definition, principles, algorithms, applications, performance evaluation, challenges, and future directions. We discuss prominent swarm intelligence algorithms, such as ant colony optimization, particle swarm optimization, and artificial bee colony algorithm, highlighting their applications in optimization, robotics, data mining, telecommunications, and other domains. Furthermore, we examine the performance evaluation and comparative studies of swarm intelligence algorithms, emphasizing the importance of metrics, comparative analysis, and case studies in assessing algorithmic effectiveness and practical applicability. Challenges facing swarm intelligence research, such as scalability, robustness, and interpretability, are identified, and potential future directions for addressing these challenges and advancing the field are outlined. In conclusion, swarm intelligence offers a versatile and effective approach to solving a wide range of optimization and decision-making problems, with applications spanning diverse domains and industries. By addressing current challenges, exploring new research directions, and embracing interdisciplinary collaborations, swarm intelligence researchers can continue to innovate and develop cutting-edge algorithms with profound implications for science, engineering, and society.
... We intended to achieve an unsupervised data-based view on putative lymphoma entities there. For this, we used the data within a swarm intelligence and self-organization framework 33,34 to visualize lymphoma based on the immunophenotypic characteristics obtained from the MLL9 data set (Fig. 6). The result is presented as a three-dimensional landscape in which elevations representing data distances, are colored as in topographic maps 35 . ...
... Only some WHO lymphoma classes were distinct within the data set, while others were not separable. Of note, we have previously shown that this unsupervised learning approach based on high dimensional biologic data corresponds well with disease entities of divergent treatment decisions 33,45,46 . Correspondingly, the clearly separable lymphoma classes HCL and CLL are treated differently from the standard lymphoma regimen (Rituximab, Cyclophosphamide, Anthracycline, Prednisolone and Vincristine). ...
Diagnostic immunophenotyping of malignant non-Hodgkin-lymphoma (NHL) by multiparameter flow cytometry (MFC) relies on highly trained physicians. Artificial intelligence (AI) systems have been proposed for this diagnostic task, often requiring more learning examples than are usually available. In contrast, Flow XAI has reduced the number of needed learning data by a factor of 100. It selects and reports diagnostically relevant cell populations and expression patterns in a discernable and clear manner so that immunophenotyping experts can understand the rationale behind the AI’s decisions.
A self-organized and unsupervised view of the complex multidimensional MFC data provides information about the immunophenotypic structures in the data. Flow XAIintegrates human expert knowledge into its decision process. It reports a self-competence estimation for each case and delivers human-understandable explanations for its decisions. Flow XAI outperformed comparable AI systems in qualitative and quantitative assessments. This self-explanatory AI system can be used for real-world AI lymphoma immunophenotyping.
... The recommended algorithm that will be tested on the non-linear OAs was devised by Thrun and Ultsch [81]. It is a swarm intelligence solver that relies on self-organization to propel the quest for an optimal clustering outcome. ...
... Even though the three-part published algorithm was developed to handle a multidimensional-scaling big-data classification problem, it is the small-and-dense data problem that is alternatively explored in this work. Regular swarm technology was not contemplated in this endeavor due to the reported issues emanating from such routines with regards to the solution accuracy in multi-objective applications and the determination of stopping criteria [81]. Clearly, by design, the preset (OA) factorial vectors are predisposed to ensure that all dataset partitions are viable in conformation to the richness property. ...
... It is the OA-induced factorial 'pre-clustering' that offers a new way to fingerprint a clustered dataset structure. As a screening/optimization exercise, it is intriguing, because the implemented bionic technology, which is borrowed to facilitate the expediting of the multi-characteristic multifactorial analysis, it actually demonstrates that a factorial group hierarchy may emerge without piloting the solver by a debatable objective function [81], but, by using a smart combination of swarm intelligence, self-organization and non-cooperative game theory [86]. It is the intrinsic properties of randomness, irreducibility and the Nash equilibrium in non-cooperative game theory in the emergent stigmergic classification solver that solidify the non-parametric (robust) annealing scheme. ...
Water scarcity is a challenging global risk. Urban wastewater treatment technologies, which utilize processes based on single-stage ultrafiltration (UF) or nanofiltration (NF), have the potential to offer lean-and-green cost-effective solutions. Robustifying the effectiveness of water treatment is a complex multidimensional characteristic problem. In this study, a non-linear Taguchi-type orthogonal-array (OA) sampler is enriched with an emergent stigmergic clustering procedure to conduct the screening/optimization of multiple UF/NF aquametric performance metrics. The stochastic solver employs the Databionic swarm intelligence routine to classify the resulting multi-response dataset. Next, a cluster separation measure, the Davies–Bouldin index, is used to evaluate input and output relationships. The self-organized bionic-classifier data-partition appropriateness is matched for signatures between the emergent stigmergic clustering memberships and the OA factorial vector sequences. To illustrate the proposed methodology, recently-published multi-response multifactorial L9(34) OA-planned experiments from two interesting UF-/NF-membrane processes are examined. In the study, seven UF-membrane process characteristics and six NF-membrane process characteristics are tested (1) in relationship to four controlling factors and (2) to synchronously evaluate individual factorial curvatures. The results are compared with other ordinary clustering methods and their performances are discussed. The unsupervised robust bionic prediction reveals that the permeate flux influences both the UF-/NF-membrane process performances. For the UF process and a three-cluster model, the Davies–Bouldin index was minimized at values of 1.89 and 1.27 for the centroid and medoid centrotypes, respectively. For the NF process and a two-cluster model, the Davies–Bouldin index was minimized for both centrotypes at values close to 0.4, which was fairly close to the self-validation value. The advantage of this proposed data-centric engineering scheme relies on its emergent and self-organized clustering capability, which retraces its appropriateness to the fractional factorial rigid structure and, hence, it may become useful for screening and optimizing small-data wastewater operating conditions.
... • Hypothesis testing (Thrun and Ultsch, 2021). Clustering can be used for hypothesis testing. ...
... Several high-quality reviews are available on the subject of measuring SOM quality [41,45,46]. In the case of normalized improvement of the manifold, the QM requires a definition of eigenvectors typically not given for non-linear DR methods (e.g., [41][42][43][45][46][47]). ...
... The artificial dataset 'Hepta' ( Figure 5) [50] provides well-defined linear separable structures and the artificial dataset 'Chainlink' (Figure 9) [50] provides well-defined linear non-separable structures. The natural dataset leukemia consists of more than d = 7000 dimensions but they are clearly separable [47,49]. For each semantic class we evaluated a selection of often-used representative QMs on projections of the above methods of artificial and real-world high-dimensional datasets with different DR methods designed for projection on two dimensions. ...
Dimensionality reduction methods can be used to project high-dimensional data into low-dimensional space. If the output space is restricted to two dimensions, the result is a scatter plot whose goal is to present insightful visualizations of distance- and density-based structures. The topological invariance of dimension indicates that the two-dimensional similarities in the scatter plot cannot coercively represent high-dimensional distances. In praxis, projections of several datasets with distance- and density-based structures show a misleading interpretation of the underlying structures. The examples outline that the evaluation of projections remains essential. Here, 19 unsupervised quality measurements (QM) are grouped into semantic classes with the aid of graph theory. We use three representative benchmark datasets to show that QMs fail to evaluate the projections of straightforward structures when common methods such as Principal Component Analysis (PCA), Uniform Manifold Approximation projection, or t-distributed stochastic neighbor embedding (t-SNE) are applied. This work shows that unsupervised QMs are biased towards assumed underlying structures. Based on insights gained from graph theory, we propose a new quality measurement called the Gabriel Classification Error (GCE). This work demonstrates that GCE can make an unbiased evaluation of projections. The GCE is accessible within the R package DR quality available on CRAN.
... These algorithms simulate the swarm intelligence (SI) concept in the social behavior of living things and are called SI-based systems. Today, these SI-based methods are widely used to solve problems such as clustering and routing in wireless sensor networks 27,28 . They have improved significantly the performance of these networks. ...
... Each element of PR new q is obtained using Eq. (28). ...
Today, wireless sensor networks (WSNs) are growing rapidly and provide a lot of comfort to human life. Due to the use of WSNs in various areas, like health care and battlefield, security is an important concern in the data transfer procedure to prevent data manipulation. Trust management is an affective scheme to solve these problems by building trust relationships between sensor nodes. In this paper, a cluster-based trusted routing technique using fire hawk optimizer called CTRF is presented to improve network security by considering the limited energy of nodes in WSNs. It includes a weighted trust mechanism (WTM) designed based on interactive behavior between sensor nodes. The main feature of this trust mechanism is to consider the exponential coefficients for the trust parameters, namely weighted reception rate, weighted redundancy rate, and energy state so that the trust level of sensor nodes is exponentially reduced or increased based on their hostile or friendly behaviors. Moreover, the proposed approach creates a fire hawk optimizer-based clustering mechanism to select cluster heads from a candidate set, which includes sensor nodes whose remaining energy and trust levels are greater than the average remaining energy and the average trust level of all network nodes, respectively. In this clustering method, a new cost function is proposed based on four objectives, including cluster head location, cluster head energy, distance from the cluster head to the base station, and cluster size. Finally, CTRF decides on inter-cluster routing paths through a trusted routing algorithm and uses these routes to transmit data from cluster heads to the base station. In the route construction process, CTRF regards various parameters such as energy of the route, quality of the route, reliability of the route, and number of hops. CTRF runs on the network simulator version 2 (NS2), and its performance is compared with other secure routing approaches with regard to energy, throughput, packet loss rate, latency, detection ratio, and accuracy. This evaluation proves the superior and successful performance of CTRF compared to other methods.
... Swarm intelligence describes the behavior of a decentralized self-organizing system consisting of elements called agents [1,2]. Swarm intelligence is based on the behavior of a community of homogeneous living creatures, e.g. ...
Swam intelligence is based on the principles of behavior of living creatures. The formalized algorithmic behavior of these creatures allows solving various mathematical problems. This paper concerns the application of swarm intelligence in solving a global optimization problem. The paper gives an example of solving a global optimization problem and presents an analysis of obtained results. This approach can be used in a variety of applications requiring the solution of global optimization problems.