Biologically inspired computing

Software systems are becoming ever more complex, as the capabilities of the software upon which they are based increase. To develop software that is manageable, we must look for novel sources of inspiration, rather than requiring an increasingly costly level of human support. Self-organising software suggests benefits in this direction, and this chapter focuses on a particularly relevant area of inspiration, that of natural systems. Indeed many natural systems are themselves self-organising, despite being often very complicated. We consider natural systems both in the context of non-living (mathematical, physical and chemical) and living (cellular, invertebrate and vertebrate, not excluding human beings) examples. We review some of the causal mechanisms and conditions that are fundamental to self-organisation in natural systems, such as: complexity, evolution, ecological interactions, animal behaviour, as well as the complexities of human behaviour, which has given us insights into phenomena such as small-world networks, epidemics, trust and gossip. In each case, we consider the relevance to self-organising software, and we throughout provide a range of references to applications and to other chapters where particular areas will be developed in more detail. We conclude by considering two application areas which arise frequently in the context of self-organising software, in different contexts: trading in markets, and networks when considered from the perspectives of design, management and robustness.

This is the readable introductory version of Global Workspace Theory, published in 1997 in the Journal of Consciousness Studies. The Oxford University Press book of that year provides an easy book-length introduction. Since 1997 our emphasis has been on incorporating more and more brain evidence. Please see our recent articles on Academia.edu

Introduction to Algorithms for Data Mining and Machine Learning (book)  introduces the essential ideas behind all key algorithms and techniques for data mining and machine learning, along with optimization techniques. Its strong formal mathematical approach, well selected examples, and practical software recommendations help readers develop confidence in their data modeling skills so they can process and interpret data for classification, clustering, curve-fitting and predictions. Masterfully balancing theory and practice, it is especially useful for those who need relevant, well explained, but not rigorous (proofs based) background theory and clear guidelines for working with big data.

The textbook is available in hard-copy version (in Serbian language) primary for our MSc and PhD students to learn and investigate modern scientific disciplines in domain of intelligent manufacturing systems and processes development. They will learn new concepts within Industry 4.0 development considering a lot of our courses like: Intelligent manufacturing systems (MSc), Industrial robots (MSc), Robotics and AI (MSc), Machine learning of intelligent robotic systems (MSc), Decision-making methods (MSc), Scheduling of manufacturing systems and processes (MSc), Autonomous systems and machine learning (PhD), Biologically inspired optimization algorithms (PhD), Systems of ANN (PhD), Cognitive robotics (PhD).

The growing complexity of real-world problems has motivated computer scientists to search for efficient problem-solving methods. Evolutionary Computation and Swarm Intelligence metaheuristics are outstanding examples that nature has been an unending source of inspiration. The behavior of bees, bacteria, glowworms, fireflies, slime molds, cockroaches, mosquitoes and other organisms have inspired swarm intelligence researchers to devise new optimization algorithms. This tutorial highlights the most recent nature-based inspirations as metaphors for swarm intelligence metaheuristics. We describe the biological behaviors from which a number of computational algorithms were developed. Also, the most recent and important applications and the main features of such metaheuristics are reported.

ABSRACT: Design methodologies have radically shifted our conception of the design process, as well as our understanding of geometrical forms as a function of performance instead of finite positions in space. However, the materials perspective that digital technologies has not yet been explored on the basis of the new possibilities disclosed by these very same tools; material considerations have almost considered as a post-rationalization design input. So far, the materialization of formal expressions instigated by such processes is primarily based on techniques of assembly, which do not negotiate the advanced levels of material complexity that the tools put forward. This lack of materialism limits the value of the design tool and the exploration of the design space, resulting in a style driven or decorative computational form making processes. This work lies on the premise of investigating modes to address an emergent rather than imposed materiality of distributions, instigated by computer-generated processes. It introduces a sustainable design approach that draws on evolutionary concepts or 'Biogenesis' of natural forms, from their material properties and from their adaptive response to changes in their environment. Instead of looking at the macro level of a building (Formal expression), this paper investigates assemblies at the domain of Material microstructures as a function of structural and environmental performance that generates design form.

Many optimization problems in science and engineering are highly nonlinear, and thus require sophisticated optimization techniques to solve. Traditional techniques such as gradient-based algorithms are mostly local search methods, and often struggle to cope with such challenging optimization problems. Recent trends tend to use nature-inspired optimization algorithms. The standard cuckoo search (CS) is an optimization algorithm based on a single cuckoo species and a single host species. This work extends the standard CS by using the successful features of the cuckoo-host co-evolution with multiple interacting species. The proposed multi-species cuckoo search (MSCS) intends to mimic the co-evolution among multiple cuckoo species that compete for the survival of the fittest. The solution vectors are encoded as position vectors. The proposed algorithm is then validated by 15 benchmark functions as well as five nonlinear, multimodal case studies in practical applications. Simulation results suggest that the proposed algorithm can be effective for finding optimal solutions and all optimal solutions are achievable in the tested cases. The results for the test benchmarks are also compared with those obtained by other methods such as the standard cuckoo search and genetic algorithm. The comparison has demonstrated the efficiency of the present algorithm. Based on numerical experiments and case studies, we can conclude that the proposed algorithm can be more efficient in most cases. Therefore, the proposed approach can be a very effective tool for solving nonlinear global optimization problems.

This work investigates the parallelization of the Artificial Bee Colony Algorithm. Besides a sequential version enhanced with local search, we compare three parallel models: master-slave, multi-hive with migrations, and hybrid hierarchical. Extensive experiments were done using three numerical benchmark functions with a high number of variables. Statistical results indicate that intensive local search improves the quality of solutions found and, thanks to the coevolution effect, the multi-population approaches obtain better quality with less computational effort. A final comparison between models was done analyzing the trade-offs between quality of solution and processing time.

This work presents an integrated simulation system for development of modular robotic locomotion patterns.
Using graphical configuration interfaces users can create robots, attach sensors, create and assign actuation algorithms, build an environment and optionally set up an optimization mode. The graphical control interface allows to interact with the simulation at runtime and to supervise it. With the possibility to run the system without graphical user interface (GUI), long-term optimizations can be performed as background processes. The system enables users to compare different locomotion patterns with the help of a standardized programming interface for actuation algorithms. Recorded data from simulations can be inspected and, if needed, exported to tab-separated value files (*.tsv) for further usage. The proposed system can be regarded as a framework for people at different levels of knowledge about modular robot locomotion principles. Its main functions are development, analysis and optimization of locomotion algorithms for different configurations of modular robots. Great benefit arises from the possibility to optimize even not well understood and complex control parameters, whose correlations with the behaviour of the associated control algorithm and other parameters are not clear. An application of the system for control parameter optimization is shown in two different experiments. In these experiments two control algorithms implemented using abstract base interfaces of the system, are automatically improved to demonstrate advantages of the system.
For the best of my knowledge the presented system offers a combination of usability, features and benefits for research in the field modular robotic locomotion that exceeds other systems.

UTS Bio Inspired Computation,
Bio Inspired Computation,
UTS Komputasi Berbasis Biologi,
Komputasi Berbasis Biologi,
Komputasi Biologi,
Bioinformatika,
Bioinformatics,
UTS Bio,
Informatika,
Teknik Informatika,
Universitas Atma Jaya Yogyakarta,
UAJY,

This paper reports the hybridization of the artificial bee colony (ABC) and a genetic algorithm (GA), ina hierarchical topology, a step ahead of a previous work. We used this parallel approach for solving theprotein structure prediction problem using the three-dimensional hydrophobic-polar model with side-chains(3DHP-SC). The proposed method was run in a parallel processing environment (Beowulf cluster), andseveral aspects of the modeling and implementation are presented and discussed. The performance of thehybrid-hierarchical ABC-GA approach was compared with a hybrid-hierarchical ABC-only approach forfour benchmark instances. Results show that the hybridization of the ABC with the GA improves the qualityof solutions caused by the coevolution effect between them and their search behavior. Copyright © 2011John Wiley & Sons, Ltd.

Lifelong learning is a long-standing aim for artificial agents that act in dynamic environments in which an agent needs to accumulate knowledge incrementally without forgetting previously learned representations. Contemporary methods for incremental learning from images are predominantly based on frame-based data recorded by conventional shutter cameras. We investigate methods for learning from data produced by event cameras and compare techniques to mitigate forgetting while learning incrementally. We propose a model that is composed of both, feature extraction and incremental learning. The feature extractor is utilized as a self-supervised sparse convolutional neural network that processes eventbased data. The incremental learner uses a habituation-based method that works in tandem with other existing techniques. Our experimental results show that the combination of different existing techniques with our proposed habituation-based method can help avoid catastrophic forgetting even more, while learning incrementally from the features provided by the extraction module.

In an attempt to provide a unified account for a vast literature discussing a multiplicity of selves, Shaun Gallagher (2013) has proposed a pattern theory of self. Subsequent discussion on this account has led to a concern that the pattern theory, as originally presented, stands as a mere list of aspects that fails to explain how they are related in real-time. We suggest that one way to address these criticisms and further develop the pattern theory of self is by exploring how it can be used to aid research on self in artificial general intelligence, especially in the context of biologically inspired cognitive architectures. We furthermore propose a conceptual implementation for actualizing such research in regards to the LIDA (Learning Intelligent Decision Agent) cognitive model.

Biocomputing is a recently growing and highly interdisciplinary field of research that investigates models and computational techniques inspired by biology and related sciences. Here, cyclic behavior (redox cycling) of purified horseradish peroxidase protein among native enzyme and its two electron-oxidized and single electron-oxidized intermediates known as Compounds I and II was algebraically expressed as a cyclic additive group Z 3 = {C 0 , C 2 , C 1 } = {C 0 , 1C 2 , 2C 2 } = {0, 2, 1}, and a cyclic multiplicative group Z * 3 = {C 1 , C 2 } = {C 1 , C 1 2 } = {1, 2}, with C 2 as the common generator. Above algebraically expressed features of the enzyme's redox cycle was applied to help determining the coefficients in polynomials formed after additive and/or multiplicative operations between polynomial rings f (x) and g(x) over a coefficient field derived from Z 3. Similarly, use of a pair of small DNA molecules was proposed for determining the coefficients for additively and multiplicatively obtained polynomials over F , (Z 2 ; +, ×); where Z 2 = {C 0 , C 1 } = {0, 1}. Discussion includes the required designs of two distinct DNA molecules for performing binary logical conjunction (AND) and exclusive disjunc-tion (XOR), upon polymerase-chain reactions.

This article reinterprets the Global Workspace theory in the form of a coherent model. The model is initial and simple. It interconnects the Working Memory and the Long Therm Memory using the graph theory.

The model can be used for software or mental experiments which can explain properties and behaviour of human mind. It is  used for explaining the magical problem of 4 and for identifying the nature of context.

This paper develops PeSeeD, a new metaheuristic algorithm for finding optimal spaced seed. Sequences matching is a hot topic in bio-informatics, which is used in many applications such as understanding the functional, structural, or evolutionary relationships between the sequences. The most relevant sequences matching methods are based on seeds designed to match two biological sequences. The first approach which introduced seeds was facilitated via Blastn tool, the approach builds seeds of 11 length size. However, it is clear that not all local alignments have to include an identical fragment of length 11. The spaced seeds approach is one of the methods which does not require a consecutive matching position. Dynamic programming is used to solve this kind of problem and it takes quadratic time. Several approaches have then been proposed to improve the sensitivity of searching in reasonable runtime. To reduce the complexity of such approaches, other heuristics based approaches can als...

Although inspired by the biological aspect, the concept of autonomic computing is emblematic of the diversity of solutions that propose a self-governance of systems from high-level objectives. From the human point of view, the autonomic nervous system has as a central point the direct intervention of a small region of the brain called the hypothalamus. This article seeks to discuss the concept of computational autonomy from an analysis of the human autonomic system, addressing its internal functioning, in order to propose a correlated computational model that can be applied. Finally, the bio-inspired computational model is mapped to an approach for knowledge-based systems aimed at its applicability.

Purpose – The purpose of this paper is to study the slime mould Physarum polycephalum as an ideal biological substrate for transport networks. When presented with several sources of nutrients the slime mould propagates colonises the sources and spans them with a network of protoplasmic tubes allegedly optimised for transfer of nutrients and metabolites. Such formation of slime mould's protoplasmic network resembles development of man-made transport systems. Thus, it sounds reasonable to compare the protoplasmic network with an established network of vehicular transport links to uncover potential (dis-)similarities between slime mould grown and man-made networks and shed more light onto general principle guiding growing biological and socio-engineering systems. Design/methodology/approach – The paper proceeds by representing major urban areas of China by oat flakes, inoculating the slime mould in Beijing, waiting till the slime mould colonises all urban areas, or colonises some and cease further propagation, and analysing the protoplasmic networks formed and comparing with man-made motorway network and planar proximity graphs. Findings – Laboratory experiments found that P. polycephalum provides a very good match for the Chinese motorway networks. Moreover, both the Chinese motorway network and the slime mould protoplasmic networks have minimum spanning trees and other proximity graphs as their sub-graphs. The experiments also identified the urban areas unlikely to be spanned by the protoplasmic networks, which may reflect hot-spots in existing challenges of modernising the motorways. Originality/value – The paper demonstrated the strong component of transport system built by slime mould of P. polycephalum on major urban areas of China consisting of one chain of four nodes and one planar graph with three leaves and eight cycles; the planar graph resides on the urban areas in the southeast part of China.

In recent years, Deep Learning techniques have shown to perform well on a large variety of problems both in Computer Vision and Natural Language Processing, reaching and often surpassing the state of the art on many tasks. The rise of deep learning is also revolutionizing the entire field of Machine Learning and Pattern Recognition pushing forward the concepts of automatic feature extraction and unsupervised learning in general.
However, despite the strong success both in science and business, deep learning has its own limitations. It is often questioned if such techniques are only some kind of brute-force statistical approaches and if they can only work in the context of High Performance Computing with tons of data. Another important question is whether they are really biologically inspired, as claimed in certain cases, and if they can scale well in terms of "intelligence".
The dissertation is focused on trying to answer these key questions in the context of Computer Vision and, in particular, Object Recognition, a task that has been heavily revolutionized by recent advances in the field. Practically speaking, these answers are based on an exhaustive comparison between two, very different, deep learning techniques on the aforementioned task: Convolutional Neural Network (CNN) and Hierarchical Temporal memory (HTM). They stand for two different approaches and points of view within the big hat of deep learning and are the best choices to understand and point out strengths and weaknesses of each of them.
CNN is considered one of the most classic and powerful supervised methods used today in machine learning and pattern recognition, especially in object recognition. CNNs are well received and accepted by the scientific community and are already deployed in large corporation like Google and Facebook for solving face recognition and image auto-tagging problems.
HTM, on the other hand, is known as a new emerging paradigm and a new meanly-unsupervised method, that is more biologically inspired. It tries to gain more insights from the computational neuroscience community in order to incorporate concepts like time, context and attention during the learning process which are typical of the human brain.
In the end, the thesis is supposed to prove that in certain cases, with a lower quantity of data, HTM can outperform CNN.

Abstract shown in the journal article.

Please note that "the frame problem" is a general challenge in AI/robotics as well as in computational neurobiology. Broadly speaking, as a robot drives outside of a familiar space, like a classroom, into a very different space, like a school hallway, it often becomes lost --- all the visual, auditory, tactile, etc. landmarks are gone. This obviously applies to simulated spaces as well a real-world spaces. Global Workspace Theory (Baars, 1988) proposed a set of Dominant Context Hierarchies to change the implicit working framework for the robot or moving animal. Such changes in Dominant Frames are  triggered by brief conscious events in the human brain. The function of the conscious event is to broadcast the new context to cortex and linked structures, thereby triggering adaptive and updating processes, such as known to exist in hippocampus, for example.

It is well known that, in nature, populations are dynamic in space and time. This means that the size of populations oscillate across their habitats over time. This work uses the concepts of habitats, ecological relationships, ecological successions and population dynamics to build a cooperative search algorithm, named ECO. This work aims to explore the population sizing not as a parameter but as a dynamic process. The Artificial Bee Colony (ABC) was used in the experiments where benchmark mathematical functions were optimized. Results were compared with ABC running alone, with and without the use of population dynamics. The ECO algorithm with population dynamics performed better than the other approaches, possibly thanks to the ecological interactions (intra and inter-habitats) that enabled the co-evolution of populations and to a more natural survival selection mechanism by the use of population dynamics.

Although inspired by the biological aspect, the concept of autonomic computing is emblematic of the diversity of solutions that propose a self-governance of systems from high-level objectives. From the human point of view, the autonomic nervous system has as a central point the direct intervention of a small region of the brain called the hypothalamus. This article seeks to discuss the concept of computational autonomy from an analysis of the human autonomic system, addressing its internal functioning, in order to propose a correlated computational model that can be applied. Finally, the bio-inspired computational model is mapped to an approach for knowledge-based systems aimed at its applicability.

One of the more interesting concepts being postulated for next gen power plants, is the notion of highly resilient and survivable I&C systems. While resilience and survivability are “cross system attributes”, resilience and robustness will rely on highly robust and fault resilient devices as a foundation. This paper presents a new self-healing digital I&C architecture inspired by the operation of two biological concepts: human immune system and embryonic stem cells. The purpose of this research is look at different bio-inspired paradigms for organizing system functions so that they become inherently resilient due to device organization and behaviors. Our current proposed approach (called BIO-SymPle) is organized in two levels: critical service layer, which is providing the intended service of critical application, and healing layer, that supervises the operation of all repair mechanisms. To facilitate coordination, organized behavior among cells we employ network-based topology stru...